Defendant and Raymond Bullock, tried jointly before us and a jury upon separate bills of indictment charging aggravated robbery, were found guilty. Counsel for Elliott moved for a new trial and in arrest of judgment, but later proceeded on the former motion only.
In the early evening of May 29, 1965, three men entered a retail cigar store operated by the Sussman family. One produced and brandished a revolver; his accomplices, intimidating the proprietors, took about *252$52 from the cash register. During the robbery, Violet Sussman was seized by her throat and her elderly father by his shirt. The thugs then fled in a car described to the police by young Mr. Sussman as a black Cadillac sedan, Pennsylvania license 66507-K.
Shortly afterward, Bullock was arrested when the police found him driving a car answering this description. Later identified by the Sussmans, he signed a written confession, implicating Elliott and one Herbert Ball. Elliott was arrested and identified both at the magistrate’s hearing and at trial.
During the course of the trial, the Commonwealth had Bullock’s confession, implicating Elliott, read into evidence, over the objection of his attorney, who had first requested expungement of all reference to Elliott. Upon the statement of the assistant district attorney that this could not be done without affecting the substance of the confession, and, we repeat, over objection, we allowed its reading, with what we considered to be clear cautionary instructions to the jury, repeated in our charge, that it was admissible against Bullock only, and was not to be considered at all against Elliott.
In essence, opposing counsel agree that the question involved is whether in a joint trial it is prejudicial error to admit into evidence a written confession of one defendant inculpating his codefendant, even though the jury is carefully instructed that the confession is not to be considered against the codefendant.
I.
We begin by recognizing that it is a long-accepted general, Federal and Pennsylvania rule of evidence that the confession is admissible when attended by proper cautionary instructions to the jury: IV Wig-more, “Evidence”, §1076; 20 Am. Jur. §493; Fife v. Commonwealth, 29 Pa. 429; Commonwealth v. Vento, 410 Pa. 350; Delli Paoli v. United States, 352 U. S. 232.
*253At the very heart of this rule lies the presumption that it must be taken for granted the jury obeys the cautionary instructions, so that the admission of what would otherwise have been legally inadmissible evidence against the codefendant is “harmless error”: Commonwealth v. Novak, 165 Pa. Superior Ct. 576.
Novak recognizes that there is a “resulting prejudice” in admitting this hearsay, inculpatory evidence, but relies upon a presumption that it is “nullified by proper cautionary instructions”, and permits the trial judge to exercise his sound discretion as to whether his instructions achieved such nullification.
To achieve sharp focus of the central issue, we must, perforce, immodestly lay claim in this case to having met and perhaps exceeded the requirements for “cautionary instructions”. Thus, we can address ourselves to the problems of prejudice to codefendant and the presumption that a jury heeds and obeys the cautionary instruction, thereby nullifying the prejudice.
The stare decisis before us, hoary and as prevalent as it is, is not without its recognized imperfections and eminent critics.
In Nash v. United States, 54 F. 2d 1006, Learned Hand referred to this rule as “. . . the recognized subterfuge of an instruction to the jury to confine its use to [the confessor] ”, and said:
“If we were to reframe the law of evidence and were still to preserve the hearsay rule, it might be better to keep out all such, for the practice, though well settled, is an evasion, and evasions are discreditable. There is no reason why the prosecution, if it chooses to indict several defendants together, should not be confined to evidence admissible against all, and if real injustice were done, the result would be undesirable. In effect, however, the rule probably furthers, rather than impedes, the search for truth, and this perhaps excuses the device ivhich satisfies form while it violates sub*254stance; that is, the recommendation to the jury of a mental gymnastic tuhich is beyond not only their powers, but anybody’s else”. (Italics supplied.)
In Krulewitch v. United States, 336 U. S. 440, 453, Justice Jackson called it a “fiction” which required a “naive assumption” about the way juries function.
Writing for the four dissenters in Delli Paoli v. United States, supra, Justice Frankfurter stated:
“The fact of the matter is that too often such admonition against misuse is intrinsically ineffective in that the effect of such a nonadmissible declaration cannot be wiped from the brains of the jurors. The admonition therefore becomes a futile collocation of words and fails of its purpose as a legal protection to defendants against whom such a declaration should not tell. . . .
“The Government should not have the windfall of having the jury be influenced by evidence against a defendant which, as a matter of law, they should not consider but which they cannot put out of their minds”: Pages247-48 (Italics supplied.)
In 24 U. Chi. L. Rev. 710 (1957), the commentator refers to the practice of instructing the jury to limit the confession to the declarant as the “placebo”, i. e., conciliative and deceptively soothing. Examining the prevalent rule and, more particularly, the holding in Delli Paoli, the observation was made that:
“It seems reasonable that a jury will follow many instructions, but it does not follow that it is reasonable to expect a jury to obey instructions to disregard relevant evidence. Research by the Jury Project at the University of Chicago Law School tends to support a widely held suspicion of trial lawyers that such an instruction only serves to make the forbidden evidence weigh more heavily in jurors’ minds, even though they may consciously attempt to follow the instruction”: Page 713.
*255This law review article favors the suggestion that severance be the price for reception of such confessions, contending that:
“Such a rule would result in separate trials only where the prosecution feels the admissions sufficiently important to justify the expense and inconvenience of separate prosecution. In any event, where post-conspiracy admissions are put in evidence at joint trial, administrative expense should not deter severance; risk of injustice would seem a heavy price for economy of administration”: Page 714.
In the same year, 1963, that our Pennsylvania Supreme Court, in Commonwealth v. Vento, supra, upheld the admission of the confession when a jury was properly cautioned, a different result was achieved in our Superior Court. In Commonwealth v. Oister, 201 Pa. Superior Ct. 251, the late Judge Flood, writing the majority opinion, said:
“The defendant Brockerman complains that his request for severance was not granted and the jury consequently heard Oister’s confession which was not admissible against him and that this was so prejudicial to him that no instruction from the court, however strong, could remove its effect from their minds”.
Continuing, Judge Flood held:
“While the evidence, exclusive of the Oister confession, may have been sufficient to convict him, it was not of such strength as to exclude the possibility that Brockerman was seriously prejudiced by the fact that the jury which convicted him heard Oister’s confession that he had hired Brockerman to set the fire and paid him for doing so. While we do not say that the trial judge was in error in refusing to sever the cases of the two defendants for trial when the application was made, in view of the result we conclude that Brockerman may well have been seriously predjudiced by the introduction of Oister’s confession during the joint *256trial, and the conviction under the circumstances cannot stand”. (Italics supplied.)
We observe that the Superior Court refrained from positively asserting that prevailing Pennsylvania law was being disregarded or overruled; that it did not categorically say it was prejudicial error beyond the cure of cautionary instructions to admit the inculpatory confession. Indeed, it did not say that Brockerman was seriously prejudiced. It reversed and ordered a new trial on its conclusion that he “may well have been seriously prejudiced by the introduction of [the] confession”.
Nonetheless, and irrespective of the absence of any ringing denunciation of the existing rule, our Superior Court did reverse, implicitly giving recognition at least to the possibility of incurable prejudice in these situations.
Of sterner stuff and more recent vintage is People v. Aranda, 47 Cal. Reporter 353, 407 P. 2d 265, decided November 12, 1965, almost two months after Elliott was tried. There, a six-to-one division resulted in California’s departure from the ranks of those jurisdictions which permitted the introduction of inculpatory confessions when attended by cautionary instructions.
After according recognition to its existing rule, and that it was shared generally and was Federal law, it called the roll of those judges and authorities who were of contrary view, many of whose utterances we have previously related, and said:
“Whether or not these criticisms of the present rule require its abrogation, a question we consider later herein, they clearly foreclose any assumption that error in admitting a confession that implicates both defendants is rendered harmless to the nonconfessing defendant by an instruction that it should not be considered against him. At best, the rule permitting *257joint trials in such cases is a compromise between the policies in favor of joint trials, and the policies underlying the exclusion of hearsay declarations against one who did not make them”.
Aranda then held, page 358:
“The giving of such instructions, however, and the fact that the confession is only an accusation against the nondeclarant and thus lacks the shattering impact of a self-incriminatory statement by him . . . preclude holding that the error of admitting the confession is always prejudicial to the nondeclarant”.
So saying, the court proceeded to reverse, because “it is reasonably probable that a result more favorable to Aranda would have been reached had [the] confession been excluded. The error therefore resulted in a miscarriage of justice”. Two people had positively identified Aranda.
Aranda, at this point in its holding, carries Oister one step further, for it moves from Judge Flood’s “possibility” of serious prejudice to the realm of “reasonably probable” prejudice.
Judicial conscience should be offended, and ours is assaulted when one reads, as in Novak (supra), that the fact of admission of a defendant’s confession is recognized as prejudicial to the nondeclarant codefendant, and but theoretically ameliorated by a judge’s instructions. This fact, we emphasize, is conscious, deliberate and intellectual, and knowingly affects the supreme values of life and liberty.
Yet we are asked to gloss over this fact as being but slightly prejudicial, assuaging our conscience and inflating our ego by the fiction that our instructions are unquestionably obeyed. In this context, a little bit of prejudice is like a “little bit pregnant” — there is no such diminutive; one is either pregnant or not pregnant; prejudice either exists or does not exist.
In civil law, strange to say, where monetary a,nd *258property values are in issue, our courts are much more sensitive to prejudice, even when accidentally introduced, and, starkly realistic, recognize the futility of corrective instructions to ignore.
Thus, in a negligence action, where evidence of the condition of a locomotive was improperly admitted, and the trial court directed the jury to “disregard the evidence and withdraw it from their consideration”, the Supreme Court said:
“It has long been held . . . that where evidence has been improperly received which tends to prejudice the minds of the jurors, . . . the (court’s) instruction comes too late and does not cure the error of admitting it . . .”. (Italics supplied.) (Note: “tends to prejudice”, obviously, is below the level of the admitted prejudice in our problem: Erie & W. V. R. Co. v. Smith, 125 Pa. 259, 264.)
Also, in an appeal from an award of viewers, it appears that there had been reference to a traction company during the property owner’s presentation. The trial court instructed the jury to disregard this. Our Supreme Court held:
“As the case is exhibited by the record we can only say that the evidence was improperly admitted, and that the instruction in regard to it was insufficient to neutralize any prejudicial effect it was calculated to produce. We are not sure that prejudice did not result to the defendant in consequence, and we sustain this assignment of error”. (Note: The court reversed, even though “it was not sure” whether predjudice resulted: Martin v. Baden Borough, 233 Pa. 452.)
Again, in a negligence action, counsel for plaintiff, closing to the jury, said: “. . .1 assure you that Mr. Lahere will not have to pay one cent of it [your verdict] ”.
“The court instructed the jury to disregard the statement by counsel for the plaintiff as to the question of payment of the verdict. It seems to us, however, that *259the remark of counsel for plaintiff connotes very clearly that the individual defendant carried liability insurance: In Kaplan v. Loev, 327 Pa. 465, 194 A. 653, the Supreme Court held: ‘The fact that the defendant in an action of trespass is insured is an irrelevant fact prejudicial to the defendant and its introduction by the plaintiff, whether by questions asked witnesses or otherwise, is reversible error’.
“In the case before us the lower court was well aware that counsel for plaintiff went beyond the limitations of a legitimate argument. The remarks of counsel, we feel, were so improper and so prejudicial that no amount of precaution taken by the court could have prevented a prejudicial effect on the jury. The court, it is true, did admonish the jury to disregard the remarks but this admonition coming after a clear innuendo as to who would pay the verdict had its effect on the jury and prevented a fair trial”: Fleet Carrier Corporation v. Lahere, 184 Pa. Superior Ct. 201, 204.
Of particular interest in this last case are (1) the impropriety was not in the testimony but in a remark of counsel, (2) the anathematic subject of “insurance” was but an implication at best and (3) instructions and admonitions were, nevertheless, recognized as incapable of “preventing a prejudicial effect on the jury”.
Thus, jurors who may alternate between civil and criminal trials are not deemed capable of obeying instructions when they ponder monetary damages, but weighing life and liberty, are! Thus, the left hand of criminal jurisprudence knoweth not what the right hand of its civil branch doeth!
We go the rest of the way opened by Oister and Aranda. We think it an impossible burden upon a trial judge, at the moment a confession is offered, to determine either the possibility or reasonable probability of prejudice, and if there is prejudice, the quantum thereof.
We hold that it is always prejudicial and was so in *260this case. We maintain that the clearest of instructions to the most intelligent of jurors does not and cannot nullify this prejudice.
We believe, with Learned Hand, that the instruction to the jury is but a “recognized subterfuge”, an “evasion” and “a mental gymnastic beyond . . . anybody’s (powers)”; with Justice Frankfurter, we hold that the effect of the confession “cannot be wiped from the brains of the jurors”. We too regard the instruction as self-deception and suspect that the University of Chicago Jury Project is correct in its finding that the instruction only serves to make the forbidden evidence weigh more heavily in jurors’ minds.
We detect no demanding jurisprudential need and, indeed, because we deem it demeaning to justice, we refuse to participate in subterfuge, unmitigated fiction, evasion, mental gymnastics and placebos. We believe that once the prejudice enters the case, instructions to the jury are about as effective as trying to remove a drop of ink from a can of milk.
II.
While not raised by either side, we see another issue in this situation; viz., is this procedure violative of due process?
Aranda approached this problem as follows:
“In Jackson v. Denno, 378 U. S. 368,. . . the United States Supreme Court held that a defendant was constitutionally entitled to have a trial judge or possibly a separate jury determine that his confession was voluntary before it was submitted to the trial jury for an assessment of its credibility. The court did not believe that a jury could separate the issue of the voluntariness of an extrajudicial statement from the issue of its truth. Tf there are lingering doubts about the sufficiency of the other evidence, does the jury unconsciously lay them to rest by resort to the confession?’ *261. ... It quoted from Justice Frankfurter’s dissent in Delli Paoli to the effect that a jury should not be permitted to be influenced by evidence against a defendant that as a matter of lato they cannot consider but as a matter of fact they cannot disregard, . . .
“Although Jackson was directly concerned with obviating any risk that a jury might rely on an unconstitutionally obtained confession in determining the defendant’s guilt, its logic extends to obviating the risks that the jury may rely on any inadmissible statements. If it is a denial of due process to rely on a jury’s presumed ability to disregard an involuntary confession, it may also be a denial of due process to rely on a jury’s presumed ability to disregard a co-defendant’s confession implicating another defendant when it is determining that defendant’s guilt or innocence.
“Indeed, the latter task may be an even more difficult one for the jury to perform than the former. Under the New York procedure, which Jackson held violated due process, the jury was only required to disregard a confession it found to be involuntary. If it made such a finding, then the confession was presumably out of the case. In joint trials, however, when the admissible confession of one defendant inculpates another defendant, the confession is never deleted from the case and the jury is expected to perform the overwhelming task of considering it in determining the guilt or innocence of the declarant and then of ignoring it in determining the guilt or innocence of any co-defendants of the declarant. A jury cannot ‘segregate evidence into separate intellectual boxes’ ”.
Astoundingly, after such clear analogical reasoning, the court refused to make the determination that this practice was constitutionally impermissible, saying it would await a holding by the United States Supreme Court that the due process clause requires such change. Instead, it retreated to safe ground, or flanked antici*262pated action, by adopting rules of practice requiring trial courts to either:
“(1) . . . permit a joint trial if all parts of the extra-judicial statements implicating any co-defendants can be and are effectively deleted without prejudice to the declarant. By effective deletions, we mean not only direct and indirect identifications of co-defendants but any statements that could be employed against nondeclarant co-defendants once their identity is otherwise established.
“ (2) . . . grant a severance of trials if the prosecution insists that it must use the extrajudicial statements and it appears that effective deletions cannot be made.
“(3) If the prosecution has successfully resisted a motion for severance and thereafter offers an extrajudicial statement implicating a co-defendant, the trial court must exclude it if effective deletions are not possible”.
We are not possessed of any such rule-making power, albeit for the mayhap brief moment at our disposal, we do have the discretion, nay, right to interpret Jackson v. Denno vis-a-vis the issue before us. We believe it applicable by its own words, by its clear rationale and by analogy, for these reasons:
(a) Justice White, for the majority, quoted Delli Paoli, but adopted the dissenting language of Justice Frankfurter, in effect reversing Delli Paoli.
(b) He also quoted this language from Krulewitch as support for the majority view:
“The naive assumption that prejudicial effects can be overcome by instructions to the jury ... , all practicing lawyers know to be unmitigated fiction”.
(c) If it is a denial of due process to rely on a jury’s presumed ability to disregard an involuntary confession, it is also a denial of due process to rely on a jury’s presumed ability to disregard a codefendant’s confes*263sion implicating another defendant when it is determining that defendant’s guilt of innocence.
We hold, therefore, that the placebo apparatus violates due process, and we draw comfort from 78 Harv. L. Rev. 211, where one reads, “Because of these close parallels between Jackson and Delli Paoli, Jackson may foreshadow a holding that the Delli Paoli procedure violates due process”.
III.
Again unexplored by prosecution or defense, but probed sua sponte by us as the final question which emerges from this record, did our trial procedure violate the sixth amendment’s guarantee of the accused’s right “to be confronted with the witnesses against him”, including the right to crossexamine such witnesses?
In Pointer v. Texas, 380 U. S. 400, the Supreme Court held that the fourteenth amendment made this guarantee applicable to the States. There, the prosecution introduced the prior testimony of a witness who had not been subject to effective erossexamination at a preliminary hearing where this testimony was taken. It was held that this encroached upon defendant’s right of confrontation.
In reaching its decision that the right of confrontation applied to trials in State courts, the court noted that the statement to the contrary in Stein v. New York, 346 U. S. 156, 195-96, could no longer be regarded as law. In Stein, it had affirmed the conviction of a defendant who had been implicated by the confessions of his codefendants, and the confessions had been admitted with an instruction to the jury to disregard them except against their respective declarants.
Aranda recognized the “grave constitutional doubts” generated by Pointer, but, as we noted, having assumed battle formation, paraded its colored flags and marched *264up the hill, it was content to simply march down again under the white flag of a new trial for defendant and a new set of rules of court.
Here again, we part company with Aranda. We look upon Pointer as an imperative which mandatorily requires us to hold the procedure we countenanced as violative of the sixth amendment.
In passing, a word about the Commonwealth’s argument that the rule “finds its reason in expediency”, and that “the Commonwealth is saved the time and expense of presenting the same evidence on separate occasions”, etc.
Aranda answered this, in note 9, as follows, at page 360:
“. . . Practical considerations of convenience must be subordinated when they run counter to the need to insure fair trials and to protect fundamental constitutional rights . . .” (citing United States Supreme Court cases).
We adopt this and the closing language of the University of Chicago Law Review article previously cited, viz.:
“. . . Administrative expense should not deter severance; risk of injustice would seem a heavy price for economy of administration”.
Reflecting, it is doubtless more expedient not to be required to provide counsel to all defendants almost at all stages; more expedient to permit the introduction of involuntary confessions; more expedient to eliminate the habeas corpus proceedings now flooding our courts; in truth, more expedient to do away with the jury system entirely. However, simply to list these expediencies is proof sufficient of Henry Ward Beecher’s aphorism that “expedients are for the hour; principles for the ages”.
Unlike some judges, we are not dismayed but applaud the recent decisions of the United States Supreme *265Court in the area of criminal procedure, and we express our condemnation of those trial judges who publicly voice their disapproval of the decisions of our highest court and their insolent determination not to pursue its mandates. These judges, assuredly, would be quick to hold in contempt anyone who dared to echo such language and attitude toward their own decisions.
In a recent speech to the New York State Bar Association, the former President of the American Bar Association, Lewis F. Powell, Jr., said (see Griswold, American Bar Association Journal, November 1965) :
“The right to a fair trial, with all that this term implies, is one of our most cherished rights. We have welcomed the increased concern by law enforcement agencies and the courts alike in safeguarding fair trial. Many of the decisions of the Supreme Court which are criticized today are likely, in the perspective of history, to be viewed as significant milestones in the ageless struggle to protect the individual from arbitrary or oppressive government”.
Erwin N. Griswold, Dean of the Harvard Law School, in an article he wrote for the November, 1965, issue of the American Bar Association Journal, called “The Long View”, discussed the law which permits introduction of past crimes to impeach a defendant’s credibility. These are his words:
“We accept much self-deception on this. We say that the evidence of the prior convictions is admissible only to impeach the defendant’s testimony, and not as evidence of the prior crimes themselves. Juries are solemnly instructed to this effect. Is there anyone who doubts what the effect of this evidence in fact is on the jury? If we know so clearly what we are actually doing, why do we pretend that we are not doing what we clearly are doing? This problem was discussed ably by Dean Wigmore many years ago. Yet very little has been done about it. Here again I fear that the states *266have yielded to inertia. It is such inertia that eventually leads to federal intervention. There is a good way for the states to avoid this intervention, and that is to take action themselves . . .
“When we take the long view, it becomes clearer that the progress we have made under the leadership of the United States Supreme Court should be accepted— indeed welcomed — and that we should all work toward continued improvement”.
We are pleased, privileged and proud to add our voice to the chorus of those who are hymning praise for the revitalization of the Bill of Rights. In what Holmes called the “path of the law”, we choose to advance, not retreat. Concluding, we quote from what Cardozo called “The Altar of Regularity”:
“Judges march at times to pitiless conclusions under the prod of a remorseless logic which is supposed to leave them no alternative. They deplore the sacrificial rite. They perform it, none the less, with averted gaze, convinced as they plunge the knife that they obey the bidding of their office. The victim is offered up to the gods of jurisprudence on the altar of regularity . . . We should know . . . that magic words and incantations are as fatal to our science as they are to any other . . . We seek to find peace of mind in the word, the formula, the ritual. The hope is an illusion”: (Cardozo: Growth of the Law — Selected Writings (Hall ed.), p. 215).
Electing not to offer up Elliott on the altar of regularity, we hold that to admit into evidence a written confession of one defendant inculpating his codefendant, even when accompanied by clear instruction to the jury that the confession is not to be considered against the codefendant, constitutes prejudicial error, violates the due process clause of the fifth amendment and the right of confrontation clause of the sixth amendment of the United States Constitution. Accordingly, we enter the following
*267Order
And now, January 18, 1966, the motion for a new trial is granted, and the motion in arrest of judgment is denied.