It is not a novel principle
that compelling a criminal defendant to appear before his or her jury in shackles without record, case specific justification, although a due process violation under the Fifth and Fourteenth Amendments to the United States Constitution, is not categorically barred from being proved harmless by the prosecution. Deck v Missouri (544 US 622 [2005]) recognizes the possibility that in hypothetically rare, extraordinarily overdetermined cases the prosecution may be able to “prove ‘beyond a reasonable doubt that the [shackling] error complained of did not contribute to the verdict obtained.’ Chapman v. California, 386 U.S. 18, 24 . . . (1967)” (id. at 635). The question now posed, then, is not what Deck in theory allows the prosecution to prove in order to save a conviction, but whether the prosecution here met its burden under Deck to demonstrate beyond a reasonable doubt that the unjustified visible shackling of this pro se defendant during his week-long jury trial—now conceded by all concerned to have been accomplished in violation of his right to due process—did not contribute to the verdict against him.
It is plain not only that no such demonstration was made, but that it was not possible given the verdict. Defendant was convicted of every count charged in the indictment. Although the evidence of some of the charged offenses was overwhelming, as the majority concedes, the evidence was not overwhelming as it bore upon the indictment’s top count, charging attempted rape in the first degree. Even if one disagrees with the conclusion shared by the trial court and Appellate Division that the evidence in proof of that count was legally insufficient,1 it is not *157arguable that the count was, at best, marginally supported. In this evidentiary context, it is apparent that it would be impossible to prove beyond a reasonable doubt that the defendant’s appearance before the jury constantly clad in officially provided implements subversive of the presumption of innocence as well as naturally and eloquently indicative of the wearer’s antisocial propensities, could not have contributed to the verdict.
Were the Court not reinstating the attempted rape verdict, as it is, it would at least be possible to argue that the shackling error did not contribute to so much of the verdict that survived. But even that argument would not in the end be analytically sound because once it is admitted that the shackling error could have infected the deliberative process, it becomes virtually impossible to prove that it affected only isolated portions of the verdict. The relevant question it must be emphasized is not whether the shackling error itself caused the jury to return the verdict it did, i.e., whether but for the error defendant would have been acquitted, but whether there is any reasonable possibility that the error contributed to the verdict. The proof of noncontribution to which the People are put under Deck is quite simply impossible where, as here, the jury has convicted with respect to any count on evidence that is less than overwhelming.
While shackling errors are not, under Deck, categorically immune from harmless error analysis but merely, as a practical matter, extraordinarily difficult to prove harmless, some shackling errors, many of which in any event would be impossible to prove harmless under Deck, are I believe sheltered from harmless error analysis under our state jurisprudence. We have long recognized that
“if in any instance, an appellate court concludes that there has been such error of a trial court, such misconduct of a prosecutor, such inadequacy of defense counsel, or such other wrong as to have operated to deny any individual defendant his fundamental right to a fair trial, the reviewing court must reverse the conviction and grant a new trial, quite without regard to any evaluation as to whether the errors contributed to the defendant’s conviction. *158The right to a fair trial is self-standing and proof of guilt, however overwhelming, can never be permitted to negate this right” (People v Crimmins, 36 NY2d 230, 238 [1975]).
“The presumption of innocence, although not articulated in the Constitution, is a basic component of a fair trial under our system of criminal justice” (Estelle v Williams, 425 US 501, 503 [1976]). Where, as here, shackles or other physical restraints imposed without adequate record justification are extensively viewed by the jury during trial and the jury is simply left to speculate upon their significance, the possibility of a fair trial becomes virtually irretrievable. The error involved, effectively depriving defendant of the presumption whose “enforcement lies at the foundation of the administration of our criminal law” (Coffin v United States, 156 US 432, 453 [1895]) was preclusive of a fair trial. While there are cases in which the brief and inadvertent exposure of a defendant in shackles to jurors will not necessarily have that consequence (see e.g. People v Harper, 47 NY2d 857 [1979]), here defendant, evidently with the Court’s imprimatur, was during his entire week-long trial made to appear before his jury visibly restrained by implements commonly and nearly inevitably understood to be reserved for the management of dangerous and explosive individuals—persons presumptively predisposed to violent crime.2 It is, or should be, clear that this appearance must have skewed the basic structure of the proceeding; the presumption of defendant’s innocence and concomitant burden of the People to prove defendant’s guilt beyond a reasonable doubt cannot reasonably be supposed to have survived. If, as Crimmins contemplates, there is a category of cases in which the fundamental, freestanding right to a fair trial must be vindicated even when there is overwhelming evidence favoring conviction, this case falls within it.
This trial was, practically speaking, hardly different from one conducted without a presumption of innocence charge or an instruction placing upon the People the burden of proving guilt beyond a reasonable doubt. When there is added to this the circumstance that the pro se defendant was literally hobbled as he exercised his Sixth Amendment right to plead his own case, it would seem clear beyond peradventure that what is at issue is *159error of a sort incompatible with the basic premises upon which trials are characterized as fair.
Accordingly, I would affirm the orders of the Appellate Division.
Judges Graffeo, Read and Smith concur with Judge Pigott; Chief Judge Lippman dissents and votes to affirm in a separate opinion in which Judges Ciparick and Jones concur.
Order, reversing County Court’s judgment, reversed, etc.
. This conclusion, it seems to me, was correct. As the trial court observed in dismissing the attempted rape count, “[i]t is undisputed that there is no mention of sex or rape, that there [was at trial no evidence of] touching of any sexual part of the victim and that there is no attempt to remove any clothing or gain access beneath any clothing.” The probative value of the DNA evidence is overstated by the majority. There was no proof as to when defendant’s semen was deposited on the leather gloves and certainly no proof that *157the deposits occurred during the charged assault. And, although there was testimony as to a “possibility” that there was a seminal fluid residue on the victim’s tee shirt, there was, apart from the admittedly indeterminate nature of the residue, no evidence as to its source.
. The majority is not wrong to say that the jury might have viewed the shackles as precautionary, but it is hard to understand how such a view would have been benign to defendant or his chances for a fair adjudication based on the fairly probative evidence.