dissenting:
The state court jury that convicted Wesley Chalmers of narcotics offenses was instructed that a reasonable doubt is “a doubt for which some good reason can be given.” The majority appears to acknowledge that this definition of reasonable doubt was error and also acknowledges that harmless error analysis is inapplicable to an erroneous definition of reasonable doubt. See Sullivan v. Louisiana, 508 U.S. 275, -, 113 S.Ct. 2078, 2081-82, 124 L.Ed.2d 182 (1993). Nevertheless, the majority holds that it was not “reasonably likely,” see Victor v. Nebraska, -U.S. —, -, 114 S.Ct. 1239, 1248, 127 L.Ed.2d 583 (1994), that the jury applied an erroneous standard of reasonable doubt and therefore rejects Chalmers’s due process challenge. In my view, the majority reaches this conclusion by failing to appreciate the significance of the trial judge’s improper definition of reasonable doubt. Because I believe it was reasonably likely — indeed, highly likely — that the instruction led the jury to apply an incorrect standard of reasonable doubt, I respectfully dissent.
I.
Instructing a jury that a reasonable doubt is “a doubt for which some good reason can be given” fundamentally misstates the reasonable doubt standard and impermissibly risks a finding of guilt by a standard less than the constitutionally required standard of proof beyond a reasonable doubt. The concept of reasonable doubt serves to place at a high level the degree of certainty that must exist in the minds of jurors before they may find a defendant guilty. See Jackson v. Virginia, 443 U.S. 307, 315, 99 S.Ct. 2781, 2787, 61 L.Ed.2d 560 (1979) (“Reasonable doubt” standard serves to “impress[ ] upon the fact-finder the need to reach a subjective state of near certitude of the guilt of the accused.”). The standard imposes no obligation whatever upon any juror to explain the juror’s thought processes to the other members of the jury. No juror need give any reason for entertaining a reasonable doubt about a defendant’s guilt, much less the “good” reason required by the trial judge’s instruction to the jury that convicted Chalmers.
The vice of such an instruction is that a juror not persuaded of guilt beyond a reasonable doubt might feel obliged to abandon a vote for acquittal because the juror feels unable to articulate to the other jurors the reason for doubting the defendant’s guilt. Though jurors are regularly instructed to listen to each other’s views, they are never placed under an obligation to articulate their views to other jurors, or even to formulate for themselves their individual reasons for not being persuaded of guilt. A juror is entitled simply to have a gut feeling that, after consideration of all the evidence, a reasonable doubt remains in the juror’s mind. Defining reasonable doubt as a doubt for which some good reason can be given tells jurors that a doubt in their minds is not a reasonable doubt unless a good reason can be given. It tells the jurors that having a reasonable doubt is not a sufficient basis on which to vote not guilty; they may vote not guilty only if the doubt in their minds is one for which they can “give[ ]” — that is, articulate — a reason. Furthermore, it tells the jurors not only that they must be able to articulate a reason for their acquittal vote, but also that their reason must be a “good” one. That message lowers the reasonable doubt standard the prosecution is required to satisfy by rejecting the adequacy of a reasonable doubt as a basis for a vote of not guilty and requiring something more.
The Court begins its process of minimizing the significance of the “good reason” instruction by acknowledging that reversible error would have resulted if the trial judge had said that “a reasonable doubt is not enough to acquit the defendant, a good reason is required.” 73 F.3d at 1268. Noting that this precise formulation was not said to the jury, the Court then expresses the view that the words that were used did not convey to the jury the very same thought. In my view, however, the message conveyed to the jury was precisely the one set forth in the Court’s hypothetical formulation. It is highly likely that every juror comes to a trial with the phrase “reasonable doubt” impressed upon the juror’s consciousness. It is a phrase *1275familiar to every person who has ever read a book about a trial, or seen a play, a movie, or a television program about a trial. When a juror hears the unusual definition “a doubt for which some good reason can be given,” it is highly likely that the juror concludes precisely what the Court’s hypothetical formulation states — a reasonable doubt alone is not sufficient to acquit; instead, a doubt suffices for acquittal only when a good reason for it can be given.
The Court resists this interpretation of the trial judge’s definition by maintaining that “[i]n context, the use of the word ‘good’ by the trial court was intended to mean doubt based on the existence or non-existence of evidence, as opposed to based on conjecture or imagination.” 73 F.3d at 1268 (emphasis added). I do not understand what contextual aspect of the jury charge permits that inference of the trial judge’s intent, but even if the Court has correctly inferred the intent of the author of the instruction, his intent is not a satisfactory basis for inferring the likely effect of the words spoken upon the listening jurors.
In any event, why are we speculating that the “some good reason can be given” definition was intended by the trial judge to convey a message that reasonable doubt is not a doubt based on conjecture, and why are we further speculating that the jurors likely understood the message as it might have been intended? It is at least as likely (and, in my view, far more likely) that the jurors understood the words to mean precisely what they said — a doubt fails to be a reasonable doubt unless the juror can give a good reason for it.
The Court also maintains that the use of the word “good” to qualify “reason” in the trial judge’s definition of reasonable doubt is similar to the jury instruction that the Supreme Court concluded passed constitutional muster in Victor. The instruction in Victor’s trial told the jury that a reasonable doubt is an “actual and substantial doubt.” — U.S. at -, 114 S.Ct. at 1249. The Supreme Court acknowledged that the adjective “substantial” was “problematic,” since it could mean either “not seeming or imaginary,” which would be “unexceptional,” or “that specified to a large degree,” which “would imply a doubt greater than required for acquittal under Winship.” Id. at -, 114 S.Ct. at 1260 (internal quotation marks omitted). The Court concluded that, in context, the adjective was “used in the sense of existence rather than magnitude of the doubt,” id., since the jury instruction expressly contrasted “substantial doubt” with “a doubt arising from mere possibility, from bare imagination, or from fanciful conjecture.” Id. at-, 114 S.Ct. at 1249.
I see little, if any, similarity to our case. In Victor, the very sentence containing the “problematic” word made clear that the word was used to mean only that a doubt sufficient to bar conviction must have “existence.” But defining the doubt in our case as one “for which some good reason can be given” does not describe the reality of the doubt; instead, the definition describes the articulation the juror is obliged to be able to offer if called upon by fellow jurors to explain what prompts the juror to entertain a doubt that the juror regards as reasonable. It is not a cheek against conjecture. It is an impermis-sibly heightened test of the circumstances under which a doubt may be regarded as reasonable. The definition increases the “magnitude” of the doubt that may bar a juror’s vote for conviction and thereby lowers the prosecution’s burden of proof, precisely the result that the Supreme Court said in Victor would violate due process.
II.
As noted, I would rule that the “good reason can be given” instruction constitutes a denial of due process, even in the absence of the five references to “moral certainty” in the prosecutor’s summation. These references, however, re-enforce my view that the instruction unconstitutionally lowered the prosecutor’s burden of proof. All five references used the bizarre phrase “to the exclusion of a moral certainty.” One appeared to link the phrase to a burden upon the defendant by stating “there’s no evidence, beyond a reasonable doubt, to the exclusion of any moral certainty, that these four men weren’t working together, because they were.” The other four references linked the phrase to the prosecutor’s burden by stating that the *1276prosecutor had “proved beyond a reasonable doubt, to the exclusion of a moral certainty” that the defendants were guilty.
The phrase “exclusion of a moral certainty” evidently comes from the prosecutor’s compounding of the customary phrase “proof that excludes a reasonable doubt” and the occasionally given, but disapproved, phrase “proof to a moral certainty.”1 The result is not merely confusing, as the majority acknowledges, it is dangerous. Taken literally (as serious arguments like jury summations are usually supposed to be), the prosecutor is telling the jury four times that he has succeeded in excluding a moral certainty, and once that the defendant has failed to exclude a moral certainty. The last reference is especially harmful, for it incorrectly implies that the defendant has a burden and then exacerbates the harm by stating that what the defendant has failed to do is to exclude a moral certainty — something the defendant obviously need not do.
But even the first four references risk harm. Surely the prosecutor was not to be understood as claiming to have excluded what he must prove, i.e., guilt beyond a reasonable doubt. The jurors would not likely think that the prosecutor was arguing how weak his case was. Instead, they would likely think he was arguing its strength, telling them that he thought that what had been excluded was the defendant’s claim of being innocent to a moral certainty. The fifth reference confirmed this likely understanding by stating it explicitly. When a jury is told that proof of guilt must exclude a reasonable doubt, and then later told that proof of guilt must exclude a moral certainty, the jury is left with the bizarre conclusion that a reasonable doubt is not just a reasonable belief of innocence, but must be a view of innocence held to a moral certainty. Thus, all five references powerfully increased the risk that the trial court’s erroneous definition of reasonable doubt left the jury with a seriously flawed understanding of the burden of proof.
If the due process requirement of proving guilt beyond a reasonable doubt is to be maintained as a meaningful requirement of constitutional law, convictions obtained after a jury instruction and a prosecutor’s summation such as occurred in this case must not be permitted to stand. I respectfully dissent.
. The original understanding of the "moral certainty” phrase, as used in the 19th century, is set forth in Victor,- U.S. at -, 114 S.Ct. at 1245-1248.