Blaine v. United States

18 A.3d 766, 2011 D.C. App. LEXIS 214, 2011 WL 1584751

• Court: District of Columbia Court of Appeals
• Decided: April 28, 2011
• Docket: 09-CF-557
• Status: Published

Opinions

FERREN, Senior Judge:

After a jury trial, Dontrace Blaine was convicted of second-degree murder while armed,¹ possessing a firearm during a crime of violence (PFCV),² and carrying a pistol without a license (CPWL).³ The court sentenced him to prison for terms totaling twenty-six years, coupled with court costs of $1,500 and followed by five years of supervised release. On appeal Blaine contends, primarily, that the trial court erred when reinstructing the jury on the government’s burden of proving guilt “beyond a reasonable doubt.”⁴ We agree with appellant that this reinstruction violated his constitutional right to due process. We therefore must reverse and remand for a new trial.

I. RELEVANT FACTS AND PROCEEDINGS

The charges against appellant grew out of a shootout in the parking lot of the Wellington Park apartment complex that resulted in the death of an innocent bystander. The government alleged that on December 29, 2006, appellant and his co-defendant, Norman Burke, had been firing at the other co-defendant, Marco Carter, when the victim was fatally struck by a stray bullet as he left his parked vehicle. Under an urban gun-battle theory,⁵ the government charged all three men with the victim’s death.

At trial, the government offered testimony from five principal witnesses, two of whom had personally observed the shooting and identified appellant as one of the shooters (both recognized him from prior dealings). In their defense, appellant and Burke each offered an alibi. Carter, in his defense, never denied that he had been present during the shootout but presented an eyewitness who testified that Carter had not possessed a gun during the incident and had ducked to avoid the bullets.⁶

After closing arguments, the trial court instructed the jury, reading the standard Redbook instruction on “reasonable doubt.”⁷ Eventually, the jury sent the trial judge a note asking for “additional guidance” on the burden of proof. Over defense objection, the trial court responded by giving a reinstruction that altered the final sentence of the Redbook instruction given before the jury retired to deliberate. Approximately two hours after reinstruction, the jury found appellant guilty on all charges.⁸

II. THE REASONABLE DOUBT INSTRUCTION

This case presents the question whether, in reinstructing on reasonable doubt in response to a note from the jury, the trial court “misdescribe[d] or lessenfed]”⁹ the government’s burden of proof and thus committed constitutional error requiring reversal of appellant’s convictions.

A. The Trial Court Decision to Reinstruct on “Reasonable Doubt”

Initially, the trial court instructed the jury on reasonable doubt with the standard, three-paragraph Redbook instruction we crafted en banc in Smith v. United States.10

The government has the burden of proving the defendant guilty beyond a reasonable doubt. In civil cases, it is only necessary to prove that a fact is more likely true than not, or, in some cases, that its truth is highly probable. In criminal cases such as this one, the government’s proof must be more powerful than that. It must be beyond a reasonable doubt.

Reasonable doubt, as the name implies, is a doubt based upon reason — a doubt for which you have a reason based upon the evidence or lack of evidence in the case. If, after careful, honest, and impartial consideration of all the evidence, you cannot say that you are firmly convinced of the defendant’s guilt then you have a reasonable doubt.

Reasonable doubt is the kind of doubt that would cause a reasonable person, after careful and thoughtful reflection, to hesitate to act in the graver or more important matters in life. However, it is not an imaginary doubt, nor a doubt based on speculation or guesswork; it is a doubt based upon- reason. The government is not required to prove guilt beyond all doubt, or to a mathematical or scientific certainty. Its burden is to prove guilt beyond a reasonable doubt.[¹¹ ]

After deliberating for more than four days, the jury sent the following note to the trial judge: “Could we please get further clarification and instruction as to the reasonable doubt standard. We have reread the instructions provided numerous times, and we would request additional guidance.” The prosecutor was skeptical: the jury had “already reread the instruction, and I don’t know that there’s a whole lot more explanation that we can provide to the jury beyond the red book explanation.” The trial judge, however, noted commentary in the Redbook that referenced a decision of this court, Payne v. United States,12 which found no plain error necessitating reversal following revisions (reflected in strikeouts and italics) to the third paragraph of our mandated instruction:

Reasonable doubt is the kind of doubt that would cause a reasonable person, after careful and thoughtful reflection, to hesitate to act in the graver or more important matters in life. However, it Reasonable doubt is not an imaginary doubt, ner[.] It is not a doubt based on guesswork or speculation or guesswork; it is a doubt based upon reason. The government -is-not-required never has to prove guilt beyond all doubt,[.] That’s impossible. They do not have to prove guilty beyond a shadow of a doubt. There’s no such thing, or They do not have to prove guilt to a mathematical m scientific certainty — and they do not have to prove guilt to a scientific certainty. Its burden is They have to prove guilt beyond a reasonable doubt.[¹³]

The trial judge proposed to reinstruct the jury by repeating Smith’s reasonable doubt instruction, augmented in paragraph three by the language from Payne. The judge explained that the Smith/Redbook instruction “is so heavily weighted to the defense, in my judgment, that an improvement, or at least a change as approved by the Payne court, is long overdue.” The government echoed the trial judge — “the original reasonable doubt instruction ... is heavily weighted toward the defense”— and then agreed with the court’s proposal. All defendants strongly objected. Counsel referenced this court’s admonition in Smith, where we stressed, “in the strongest terms, that the trial court should ‘resist the temptation to stray from, or embellish upon, that instruction.’ ”¹⁴ Furthermore, they noted, the Payne court, a three-judge division, had no authority to modify the en banc mandate.¹⁵ The trial judge acknowledged that defense counsel had made “a point,” agreeing that the reinstruction “is in one direction here, and that may be a problem.” A colloquy then ensued in which defense counsel convinced the judge to omit two brief sentences from Payne: “That’s impossible” and “There’s no such thing.”¹⁶ Even with those omissions, however, counsel argued that the cumulative impact of the revised instruction would amount to “cheerleading for the government without any sort of really substantial clarification of the standard.” The judge (while acknowledging that his “reexplaining” would be “more graphic”) was not moved. Finally, counsel offered language to counterbalance the additional language from Payne — again, without success.¹⁷

The judge then brought the jurors into the courtroom and answered their note, first, by saying: “I’m going to ... give you an instruction now that is much like the reasonable doubt instruction originally given, but with some change that may be helpful.” Whereupon he read the first two paragraphs of our en banc Smith instruction, followed by his revised Payne instruction.

Reasonable doubt is the kind of doubt that would cause a reasonable person, after careful and thoughtful reflection, to hesitate to act in the graver or more important matters in life. However, it is not an imaginary doubt, nor a doubt based on speculation or guesswork; it is a doubt based upon reason. The government never has to prove guilt beyond all doubt, they do not have to prove guilt beyond a shadow of a doubt, they do not have to prove guilt to a mathematical certainty, and they do not have to prove guilt to a scientific certainty; they have to prove guilt beyond a reasonable doubt. (Emphasis added.)

B. Appellant’s Contentions

Appellant contends that the reinstruction violated his right to constitutional due process. As orally conveyed by the trial judge, he says, the revised third paragraph, incorporating language from Payne,18 added heft to the government’s case at the expense of the defense. It transmuted the carefully balanced Smith instruction into an unbalanced one in the government’s favor amounting to a virtual invitation to convict.

In the first place, in response to the jurors’ note, the trial judge told them that he would give them “an instruction now that is much like the reasonable doubt instruction originally given, but with some change that may be helpful.” (Emphasis added.) The jurors then heard again the first two paragraphs of the Smith instruction, followed by the new language in which the judge had told them to look for “change”- — for something different.

Next, appellant points to three embellishments of Smith’s third paragraph.¹⁹

(1) In Smith, the government is “not” required “to prove guilt beyond all doubt.” In the trial court’s reinstruction, the government “never” has to do so.

(2) In Smith, the government does “not” have to prove guilt “beyond all doubt.” In the reinstruction, the government “never” has to prove guilt “beyond all doubt” and does not have to prove guilt “beyond a shadow of a doubt.”

(3) In Smith, the government is not required to prove guilt “to a mathematical or scientific certainty.” In the reinstruction, “[t]hey do not have to prove guilt to a mathematical certainty, and they do not have to prove guilt to a scientific certainty.”

Appellant does not dispute the language of reinstruction. As stated in his reply brief: “Blaine has never argued that the reinstruction was erroneous because it contained an incorrect statement of the law.” Rather, he says, the problem with the reinstruction is “the context in which it was given.” When prefaced by the judge’s invitation, after more than four days of deliberation, to look for “some change,” the reinstruction “told the jury not to hold the government to too high a standard.” By adding new as well as emphatic and repetitive language reinstructing as to what reasonable doubt is “not,” says appellant, the court created a reasonable likelihood that the jury was misled about the meaning of reasonable doubt, even though the language of reinstruction, if used as the initial charge to the jury, might well have survived constitutional challenge. The hammer-like language of reinstruction, claims appellant, became language of advocacy for the government.

Focusing more specifically on language, appellant stresses that the jurors would surely have picked up on the addition to the instruction of an entirely new formulation, “beyond a shadow of a doubt.” That new “shadow” formulation, he argues, injected ambiguity from which at least two reasonable understandings were possible. Some jurors might have equated “beyond a shadow of a doubt” with “beyond all doubt” in the original instruction. Even so, he maintains, that redundancy would have “compounded] the statements deem-phasizing the government’s burden.” Other jurors, however, might have found substantive change in the language, conceptualizing three levels of doubt: “beyond all doubt,” “beyond a shadow of a doubt,” and “beyond a reasonable doubt.” According to appellant, that new conceptualization would have dropped “reasonable doubt” to a lower level than these jurors would have perceived in Smith’s two-level distinction between “all doubt” and “reasonable doubt.” Either way, stresses appellant, when coupled with the trial court’s reference to “change” in an instruction “much like” the first one (but thus not completely so), there is a reasonable likelihood that the jury perceived a drop in the level of doubt required for conviction.²⁰

Furthermore, argues appellant, when the trial judge read the reinstruction, the jurors heard greater emphasis in the new introductory word “never ” having to prove guilt beyond all doubt, rather than merely “not” having to do so. This new emphasis tended to encourage, however subtly, the perception of a substantive distinction between “beyond all doubt” and the “beyond a shadow” clause that followed it. Finally, when the judge changed the instruction from “mathematical or scientific certainty”²¹ in one clause to “mathematical certainty” and “scientific certainty”²² in two clauses, the jurors heard repetition in parallel form. This kind of repetition, argues appellant, not only increased the emphasis on what reasonable doubt is “not,” but also created a “mantra like” cadence of “not” words that pressed the point even more powerfully than mere repetition itself.

As a result of trial judge comment and language change, concludes appellant, the judge gave the jurors an “unbalanced” instruction, the very danger from reinstruction that we warned against in Davis v. United States.²³ From our short sentence in Smith — “The government is not required to prove guilt beyond all doubt, or to a mathematical or scientific certainty”— the trial court shifted to a much longer sentence that reflects a change from one “not” to three “nots” and a “never,” in order to explain the level of doubt the jurors need not have:

The government never has to prove guilt beyond all doubt, they do not have to prove guilt beyond a shadow of a doubt, they do not have to prove guilt to a mathematical certainty, and they do not have to prove guilt to a scientific certainty[.] (Emphasis added.)

Therefore, argues appellant, the jurors — invited to be alert to “change”— could well have heard that final sentence, unlike the language of Smith, to say: they merely have to prove guilt beyond a reasonable doubt, a result not as difficult to come by as the jurors thought before. And this, appellant stresses, was no accident. The trial judge himself recognized that he was adding his thumb to the scale in the government’s favor by creating a reinstruction “more balanced” than our en banc language in Smith, which “is so heavily weighted to the defense.” The judge agreed with defense counsel that the reinstruction language went “in one direction,” and that this “may be a problem.” In light of his intention, he had to know from long experience that the jurors were likely to find the second instruction materially different from the first. The “change” the judge offered the jurors, therefore, was not merely in derogation of the carefully balanced instruction mandated by Smith, argues appellant; it was a violation of due process that must lead, ineluctably, to reversal

C. The Government’s Responses

As to appellant’s last point, the trial judge’s intention, the government responds that the judge “did not design the supplemental instruction to lower the government’s burden of proof.” When the judge characterized the en banc instruction as “heavily weighted toward the defense,” he was merely looking toward a “more balanced way” of describing reasonable doubt accurately to the jury.

Addressing the language of reinstruction, the government replies that appellant has failed to “demonstrate that the supplemental instruction was an incorrect statement of the law”; the trial court did not deviate from Smith “in any substantial way.” Appellant, as we have noted, concedes the point. The government, however, reinforces its response by stressing that legal correctness is all that is required for reinstruction, and that this court — as the trial court recognized — had approved in Payne the very embellishments of Smith at issue here. More specifically, says the government, the first and third changes appellant cites (the use of “never” and the lengthier references to “mathematical” and “scientific” certainty) were taken directly from Payne and are merely “stylistic.” The middle addition to the reinstruction (“beyond a shadow of a doubt”) was, as this court said in Payne, linguistically equivalent to the formulation immediately preceding it (“beyond all doubt”). Both, we said, mean “need not prove to a certainty.”²⁴

Next, the government rejects as “considerably exaggerate^]” appellant’s “contextual” argument that adds legal significance to the new, “shadow” language and words of emphasis and repetition. The government minimizes that argument because the reinstruction included “nonsubstantive changes in what remained a single sentence.” (Emphasis in government’s brief.) Furthermore, the government reminds us, the trial court removed “the most argumentative phrases in the Payne instruction: ‘that’s impossible,’ and ‘there’s no such thing.’ ”²⁵

Finally, agreeing with appellant that any reinstruction must be “fairly balanced,”²⁶ the government finds no disqualifying “imbalance[]” in the reinstruction. It contends that this court’s concern about lack of balance typically refers to a trial court’s emphasis in reinstruction on one aspect of the case, such as the elements of voluntary manslaughter, to the exclusion of another key aspect, such as the criteria for self-defense.²⁷ The government appears to acknowledge, nonetheless, that the court’s reinstruction on reasonable doubt could be unbalanced if it “had included only the sentence appellant complains of.” But, says the government, the court repeated the entire reasonable doubt instruction, with “six sentences” describing what reasonable doubt is and “only two sentences” spelling out what it is not. Accordingly, whatever added language there may have been for appellant to complain about in the reinstruction, it was more than offset by language from the original instruction, reread to the jurors, that reinforced the language favorable to the defense.

III. ANALYSIS

A. Standard of Review

We consider first our standard of review. “Decisions regarding reinstruction of a jury are committed to the discretion of the trial court; absent abuse of that discretion we will not reverse.”²⁸ Review for abuse of discretion, of course, has two components: whether the trial judge erred and, if so, whether the error was of a “magnitude to require reversal.”²⁹ As to error, the Supreme Court has held that the test for constitutional error in evaluating instructions on reasonable doubt is whether there is a “reasonable likelihood that the jurors who determined ... guilt applied the instructions in a way that violated the Constitution[,]”³⁰ a test this court had applied earlier.³¹ This standard of review applies to all constitutional challenges to jury instructions, whether the reviewing court focuses exclusively on the language of an initial instruction,³² addresses alleged inconsistency between two initial instructions,³³ or evaluates the interplay between instruction and reinstruction.³⁴

When the reviewing court concludes that the trial court erred while instructing the jury, the question whether the error requires reversal ordinarily invokes harmless error analysis under Kotteakos³⁵ (non-constitutional error) or Chapman³⁶ (constitutional error). Some constitutional errors, however, are not the usual kinds of “trial error” that “occur[ ] during the presentation of the case to the jury,” which may be “quantitatively assessed in the context of other evidence presented in order to determine whether [their] admission was harmless beyond a reasonable doubt.”³⁷ This other category of errors reflects “structural defects in the constitution of the trial mechanism, which defy analysis by ‘harmless-error’ standards[,]” such as the deprivation of counsel at trial or a biased trial judge.³⁸ Structural errors, therefore, are reversible automatically because they “affect[ ] the framework within which the trial proceeds, rather than simply an error in the trial process itself.”³⁹

In Sullivan v. Louisiana,⁴⁰ the Supreme Court held that a constitutionally deficient instruction on reasonable doubt is structural error. The “essential connection to a ‘beyond a reasonable doubt’ factual finding cannot be made where the instructional error consists of a misdescription of the burden of proof, which vitiates all the jury’s findings.”⁴¹ Simply put, therefore, a trial that yields a conviction based on an unconstitutional burden of proof is no trial at all, and thus the instructional error cannot be harmless.⁴²

B. The Risk Inherent in Reinstruction

Before addressing the reinstruction at issue here, we believe it is important to emphasize the substantial risk inherent in any reinstruction. When a jury, during deliberations, requests and receives a supplemental instruction from the trial judge, the risk of error is heightened because “a supplemental instruction ‘will enjoy special prominence in the minds of the jurors.’ ”⁴³ Accordingly, we emphasized in Davis v. United States that “the trial judge must be especially alert not to send the jury back to resume deliberations having most recently heard supplemental instructions which are unbalanced.... Because the ‘last word is apt to be the decisive word,’ the trial judge must prevent the ‘poi-sonfing of] an otherwise healthy trial’ by improperly balanced supplemental instructions.” ⁴⁴ Therefore — of importance here — this is not merely a language case; it is a supplemental language case.

The risk is particularly high from supplemental language explaining reasonable doubt. In Smith we stressed, “in the strongest terms, that the trial court should ‘resist the temptation to stray from, or embellish upon, that instruction.’ ”⁴⁵ We were concerned, primarily, that any deviation would risk constitutional error and, as a result, automatic reversal of a conviction.⁴⁶ It is highly questionable, therefore, for any trial judge to deviate from Smith’s exact language. As we said in Foreman v. United States with reference to reasonable doubt: an “instruction central to the determination of guilt or innocence may be fatally tainted by even a minor variation which tends to create ambiguity.”⁴⁷

To be clear: in Smith this court, sitting en banc, prescribed language describing reasonable doubt that we believe accurately reflects the constitutional standard- — and thus the minimum burden of proof required for a finding of guilt — that trial judges must convey to the jury in criminal cases. Any reinstruction that allows a jury to find reasonable doubt based on a lower level of proof will be improper,⁴⁸ and reversal will be required if this court concludes that, as a result of that reinstruction, there is “ ‘a reasonable likelihood that the jury understood the instructions to allow conviction based on a lesser standard than proof beyond a reasonable doubt’ ” as defined in Smith.⁴⁹

There is a related concern. In addition to the particularly high risk of constitutional error from merely revising the language of reasonable doubt used for rein-struction, there is a correspondingly high risk that, if the reinstruction appears to favor the government, the jury will get the message that the judge believes the defendant is guilty and thus that the government has met its burden of proof. The jurors, in asking for clarification, presumably will be listening carefully for any change in the judge’s words of guidance, and they will probably be even more alert to change if the judge, as in this case, states expressly that it’s “change” they will hear. Change in language, however, as appellant concedes, does not necessarily imply change in meaning. But even if a judge’s invitation to look for “change” will lead many, if not most, jurors to expect at least some new meaning from reinstruction, mere change as such does not necessarily suggest whom that new meaning is likely to favor. The critical concern, therefore, is the trial court’s presentation overall: whether there is a reasonable likelihood that the jurors will have heard the judge, through all aspects of reinstruction, take the government’s side by inviting them to find guilt more easily than they could have under the reasonable doubt language they heard, initially, in the judge’s recitation from Smith.

C. The Reinstruction at Issue

We turn now to the reinstruction itself. For purposes of this case, appellant does not dispute that the reinstruction is a correct statement of the law when scrutinized literally, without regard to how the trial court presented it or to the impact of comparing it with the initial instruction. Solely for the sake of argument, therefore, we assume that the Payne language, when added to the Smith instruction, would not — without more — reflect constitutional error.⁵⁰

1. The Language

We look, initially, at the language of the reinstruction and the relevance of this court’s decision in Payne that found no plain error when the trial court used similar language. The government takes the position that this court has embraced Payne: end of case. According to the government, “although Payne did arise on plain-error review, this Court was quite emphatic that it did not see anything erroneous at all in the complained-of language .... Thus, [Payne ] is not a case in which the Court found error, but no prejudice, or even expressed doubt or ambivalence about the question of legal error.” The government reminds us that in the only place where the trial court arguably inserted new substantive language from Payne in the reinstruction — namely, the addition of “beyond a shadow of a doubt” — that expression meant the same thing as “beyond all doubt” in the initial instruction. The government therefore stresses that these two formulations are redundant; in Payne, this court said expressly that both mean “need not prove to a certainty.”⁵¹

This argument is unpersuasive. First, because Payne reviewed for plain error (there was no objection to the instruction at trial), that decision cannot be said to have ruled definitively that, in a reasonable doubt instruction, all jurors under all circumstances must be presumed to understand that “beyond all doubt” and “beyond a shadow of a doubt” mean the same thing.⁵² Second, Payne did not involve the circumstance of reinstruction, with the additional risk of error that reinstruction invites. And that additional risk, as we have explained, is intensified by the trial court’s comment that the jurors should expect to find “change” in that language. Payne, therefore, does not resolve this case.

2. The Context: Tnal Court Invitation to Find “Change" in Reinstruction

That brings us to appellant’s central contention: In offering the jury the new third paragraph of reinstruction to compare with the corresponding paragraph of the initial instruction — an offer preceded by an invitation to find “change” — there is a reasonable likelihood that the jury came to an understanding that impermissibly lowered the burden of proof. We agree.

In the first place, when listening for “change” in the reinstruction, in comparison with the initial instruction, there is a reasonable likelihood that jurors perceived new substance in the judge’s addition of the “shadow” language — language that cut in the government’s favor by ostensibly creating three, no longer two, levels of doubt (as elaborated in appellant’s contentions summarized earlier).⁵³ Second, the reinstruction became unbalanced from added weight on the government’s side created by an extended rat-a-tat explaining what reasonable doubt is “not.” The court’s use of new, legally correct though “more graphic,” emphatic, and repetitive language appeared to lighten the government’s burden of persuasion.⁵⁴ Therefore, even if the language of reinstruction itself was not inherently a violation of due process, there is a reasonable likelihood that the judge’s second instruction conveyed to the jury a lower standard of reasonable doubt than due process requires, and that the jury came to its verdict accordingly.⁵⁵

The government’s reply to this contextual argument stresses: (1) that the rein-struction contained “nonsubstantive changes in what remained a single sentence ” (emphasis in government’s brief), and (2) that the reinstruction was balanced because the trial court had repeated all three paragraphs of the instruction, with “six sentences” describing what reasonable doubt is and “only two sentences” spelling out what it is not. The first argument misses the point because it altogether fails to take into account the particular impact of the modified Payne language as rein-struction. We may be dealing with a single sentence, but, in contrast with the penultimate, eighteen-word sentence in the Smith instruction, the new replacement sentence in the reinstruction is long and forceful and includes several additional “nots” (or the equivalent). In the context here, after more than four days of jury deliberation, there is a reasonable likelihood that the new language had an effect akin to that of the controversial “dynamite” or anti-deadlock charge used to prod an apparently deadlocked jury to come to unanimity.⁵⁶ After reinstruction, the jury took only two more hours to arrive at its guilty verdict.

The government’s other argument — that the reinstruction is not unbalanced because six sentences explain what reasonable doubt “is” while only two sentences say what it is “not” — is considerably overstated and thus not convincing. The first three sentences, comprising the first paragraph of the instruction, are devoted to explaining the difference in burden of proof in civil and criminal cases, presumably to alert jurors who may have served in civil cases that the burden in criminal proceedings is greater: “reasonable doubt,” not “more likely true than not” or “highly probable.”⁵⁷ We cannot discern why those sentences should be counted as instruction about what reasonable doubt “is” when they offer no specifics about what reasonable doubt itself means.⁵⁸ When we consider the three other sentences on which the government relies for balance, an express statement of what reasonable doubt “is” appears only twice (coupled with one followup, explanatory sentence).⁵⁹ In the reinstruction sentence on which appellant relies, however, describing what reasonable doubt does not include, there are three “nots” and a “never.”⁶⁰ (The sentence immediately preceding that one includes two more clauses stating what reasonable doubt is not, adding both a “not” and a “nor.”)⁶¹ The government, therefore, has not convinced us that the reinstruction, in context, was balanced.

3. Conclusion

We confirmed in Smith that some instruction- — a standard instruction — addressing “reasonable doubt” is necessary.⁶² We warned that tinkering -with the Smith instruction is so likely to lead to unbalanced language that trial judges should not do so when instructing the jury at trial’s end.⁶³ And we now conclude, in the same “strongest terms” expressed in Smith, that trial judges should “resist the temptation to stray from, or embellish upon, that instruction”⁶⁴ when confronted by a note from the jury seeking further guidance on “reasonable doubt.” We will not say that there can never be circumstances when reinstruction on reasonable doubt might survive appellate challenge, but the risk of reversal will be so great that trial judges should avoid doing so.⁶⁵

In this case, the jurors had gone over the original instruction, which they had in writing, to the point that they surely focused on the new material for clues as to what, more specifically, reasonable doubt really means. As we have recognized,⁶⁶ the new language would have been the “freshest” in the jurors minds; they would have given it “heightened alertness”; and they would have accorded it “special emphasis” — all because they had heard the new words during a break in their deliberations granted to answer a question the jurors themselves deemed important, indeed dispositive. Thus, they were vulnerable to finding conclusive meaning in anything noticeably new — new content in the “shadow” language, and new warnings about what reasonable doubt is “not” — in contrast with language in the initial instruction which they had read over and over.

But jury vulnerability was not the only dynamic at work here. The record makes clear that the experienced trial judge anticipated that the reinstruction would create a revised understanding of “reasonable doubt.” He found the Smith instruction “so heavily weighted to the defense ... that an improvement, or at least a change as approved by the Payne court is long overdue” — a change that the prosecutor, after initial skepticism, encouraged. The judge took Payne, a product of plain error review that did nothing to revise our en banc ruling in Smith, and used it to reba-lance, in the government’s favor, the very instruction we had announced definitively — and the judge himself had given initially — -as the balanced instruction required. The result was a reinstruction with changed language that, as the judge put it, went “in one direction” — in the government’s — which “may be problem.” The jury’s relatively quick verdict after rein-struction, two hours for conviction after more than four deliberative days before reinstruction, is telling.

We noted earlier that the impact of the reinstruction, under these circumstances, was akin to a “dynamite” or anti-deadlock charge.⁶⁷ As applied to the reasonable doubt instruction, that impact was particularly serious because, however individual jurors would parse the reinstruction, there was a reasonable likelihood that, collectively, the jurors would gain the overall — and correct — impression that the trial judge was restating the instruction in the government’s favor, and thus that the concept of “reasonable doubt” was less stringent than they originally had thought. Once that impression is conveyed, it is virtually tantamount to an indication that the judge believes the defendant is guilty and that the government, therefore, has met its burden of proof.⁶⁸

We conclude, accordingly, that the two instructions before us here, each of which we assume, for sake of argument, would survive constitutional challenge as an initial instruction on reasonable doubt, conveyed different meanings when the trial judge, for purposes of reinstruction, not only reworked critical language but also told the jurors to see the instructions in the new light of “some change.” Even if a grammarian, in the quiet of a study, could discern no legal difference between the message conveyed in the third paragraph of the Smith instruction and the message from Payne offered days later in place of it, we are satisfied that, in the context of reinstruction and the proceedings taken as a whole, “there is a reasonable likelihood” that the jury applied the challenged rein-struction “in a way that violate[d] the constitution” ⁶⁹ — that is, in a way that dropped the level of doubt essential for conviction below the level required by due process.

In announcing this conclusion we must return, briefly, to our standard of review. Earlier we noted the Supreme Court’s ruling in Stdlivan that, when the reasonable doubt instruction “consists of a misdescription of the burden of proof,” that error is “structural,” requiring automatic reversal, because the error “vitiates all the jury’s findings.”⁷⁰ Sullivan considered instructional language alone in finding misdescription of reasonable doubt. The present case is different. We have found misdescription from a reinstruction with language presumed constitutional in isolation but written to favor the government when compared, in the trial judge’s words, to the “reasonable doubt instruction originally given” and conveyed to the jury with a judicial invitation to find “some change.”

This particular combination of instructional language and trial court comment created a misdescription of reasonable doubt for two reasons. First, the judge’s invitation to find “some change” provided an interpretative nexus between the first and second instructions; it was inherent in reinstruction. Second, the new language and the judge’s comment, taken together, created “a reasonable likelihood that the jurors ... applied the instructions in a way that violated the Constitution” by relaxing too far them understanding of reasonable doubt.⁷¹ We see no principled basis for concluding that the universe we recognize for misdescription of reasonable doubt (comment and instruction) should be treated differently from the narrower universe grounding structural error, as in Sullivan, solely on the formal words of reinstruction. In both situations, the jurors are led to an unconstitutional lowering of the standard for reasonable doubt.

On the other hand, we recognize that once the basis for structural error extends beyond the formal words of instruction or reinstruction, the rationale for applying structural rather than harmless error analysis can become attenuated. A judge’s actions in connection with reinstruction can generate alternative theories of analysis, as our concurring colleague demonstrates. Furthermore, the Supreme Court’s 6 to 3 decision in Neder v. United States⁷² reveals the Court’s difficulty in distinguishing between structural and harmless error. Finally, whichever analysis of error is applied in this case, reversal is required because the process of rein-struction resulted in a description of the burden of proof that created a reasonable likelihood of the jurors’ applying a compromised standard of reasonable doubt. Accordingly, we need not rule definitively whether the instructional process here produced structural or harmful error. However the error is characterized, appellant’s convictions must be reversed and the case remanded for a new trial.⁷³

So ordered.

1 . D.C.Code §§ 22-2103, -4502 (2001).

2 . D.C.Code § 22-4504(b) (2001).

3 . D.C.Code § 22-4504(a) (2001).

4 . Appellant also maintains that die trial court abused its discretion by (1) refusing to sever his trial from that of his co-defendants; (2) failing to strike the testimony of a witness who conferred with the prosecutors during a break in her testimony; (3) failing to question a juror about potential bias revealed while the jury was deliberating; and (4) declining to strike testimony about appellant’s flight from law enforcement officers as inadmissible hearsay, and permitting the jury to consider that flight as evidence of guilt.

The jury acquitted one of appellant’s co-defendants on the murder and other principal charges and hung on all the charges against the other co-defendant. Because the government, according to its brief, has not sought to retry either co-defendant, the severance issue is not likely to recur upon appellant’s retrial. Also, the alleged errors concerning the prosecutor’s mid-trial contact with a witness, and the court's failure to explore juror bias, are not likely to recur. Finally, because the government argues only harmless error to defend the flight testimony, we assume that in the event of a retrial any flight evidence, if presented, will not suffer the same defect. Accordingly, we see no need to resolve these four issues.

5 . See Roy v. United States, 871 A.2d 498, 505-09 (D.C.2005).

6 . Carter also called a crime scene analyst, who testified that no shots had been fired from the area of the parking lot where Carter was located.

7 . Criminal Jury Instructions for the District of Columbia, No. 2.108 (5th ed. rev. 2010) (the "Redbook”).

8 .The jury acquitted Carter of second-degree murder while armed, voluntary manslaughter while armed (as a lesser included offense), and PFCV but remained undecided on the other charges against him: CPWL and obstruction of justice. The jury also remained undecided on all of the charges against Burke: second-degree murder while armed, voluntary manslaughter while armed (as a lesser included offense), PFCV, and CPWL. After three more days of deliberation, the jury still could not reach verdicts on the charges remaining against Carter and Burke, and the court declared a mistrial as to all of them. According to its brief, the government has not sought to retry Carter or Burke.

9 . Smith v. United States, 709 A.2d 78, 81 (D.C.1998) (en banc).

10 . Id. The Redbook instruction at the time of trial, see supra note 7, contained the Smith instruction verbatim.

11 . Id. at 82.

12 . 932 A.2d 1095 (D.C.2007).

13 . Compare id. at 1102 with Smith, supra note 9, 709 A.2d at 82.

14 . Smith, supra note 9, 709 A.2d at 83 (citation omitted).

15 . See M.A.P. v. Ryan, 285 A.2d 310, 312 (D.C.1971) (“[N]o division of this court will overrule a prior decision of this courl[;] ... such result can only be accomplished by this court en banc.”) (footnote omitted).

16 . See text accompanying supra note 13.

17 . Defendants asked for language that proof beyond a reasonable doubt “means proof that eliminates any doubt that a reasonable person might draw from a careful consideration of the evidence and lack of evidence in this case.” In its brief, the government correctly points out that this language would have been inappropriate "because it suggested that reasonable doubt could be 'any’ doubt, no matter how slight.”

18 . See supra Part II. A. (end).

19 . Compare text accompanying supra note 11 with supra Part II. A. (end).

20 . How else, appellant asks, can one explain why the jury deliberated for more than four days in indecision, followed by only two hours after reinstruction to agree unanimously on appellant’s guilt?

21 . "The government is not required to prove guilt ... to a mathematical or scientific certainty."

22 . "They do not have to prove guilt to a mathematical certainty, and they do not have to prove guilt to a scientific certainty[.]”

23 . 510 A.2d 1051, 1053 (D.C.1986) (per curiam) (affirming manslaughter and weapons convictions, court found no abuse of discretion in trial judge's refusal to include in rein-struction a reminder that "self-defense is a legal excuse to a charge of homicide,” although "a reinstruction on self-defense would have come closer to the ideal of a neutral, balanced instruction”); accord, Yelverton v. United States, 904 A.2d 383, 389 (D.C.2006) (noting "special care must be taken to assure that the instruction is fairly balanced” while holding harmless the trial court's erroneous, unbalanced reinstruction that addressed only the portion of a police officer's testimony that favored government).

24 . Payne, supra note 12, 932 A.2d at 1102.

25 . See text supra at note 13.

26 . Yelverton, supra note 23, 904 A.2d at 389.

27 . See Coreas v. United States, 565 A.2d 594, 599 (D.C.1989); Davis, supra note 23, 510 A.2d at 1052-53; see also Bates v. United States, 834 A.2d 85, 94 (D.C.2003) (no plain error in court's failure to reinstruct on "deliberation” when reinstructing on "premeditation” in first-degree murder case, because jurors had received full instructions in writing).

28 . Davis, supra note 23, 510 A.2d at 1052.

29 . Johnson v. United States, 398 A.2d 354, 366 (D.C.1979).

30 . Victor v. Nebraska, 511 U.S. 1, 22-23, 114 S.Ct. 1239, 127 L.Ed.2d 583 (1994). The Court in Victor noted, first, that the “government must prove beyond a reasonable doubt every element of a charged offense[,] ... a requirement of due process[.]” Id. at 5, 114 S.Ct. 1239 (citing In re Winship, 397 U.S. 358, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970)). The Court then concluded that, in reviewing a challenge to a reasonable doubt instruction, the court must determine "whether there is a reasonable likelihood that the jury understood the instructions to allow conviction based on proof insufficient to meet the Winship standard.” Id. at 6, 114 S.Ct. 1239 (citing Estelle v. McGuire, 502 U.S. 62, 72, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991) ("in reviewing an ambiguous instruction such as the one at issue here [concerning admission of prior injury evidence in prosecution for murder of an infant child], we inquire ‘whether there is a reasonable likelihood that the juiy has applied the challenged instruction in a way' that violates the Constitution") (citation omitted)).

31 . Foreman v. United States, 633 A.2d 792, 794 (D.C.1993) (quoting United States v. Merlos, 299 U.S.App. D.C. 401, 403, 984 F.2d 1239, 1241 (1993)) (quoting Estelle, supra note 30, 502 U.S. at 72, 112 S.Ct. 475); accord, Proctor v. United States, 685 A.2d 735, 741 (D.C.1996) ("the combined changes to the standard definition [of reasonable doubt] here created 'a reasonable likelihood that the jury understood the instructions to allow conviction based on a lesser standard than proof beyond a reasonable doubt' " (citing Foreman and quoting Butler v. United States, 646 A.2d 331, 334 (D.C.1994), cert. denied, 514 U.S. 1009, 115 S.Ct. 1326, 131 L.Ed.2d 206 (1995))).

32 . See, e.g., Victor, supra note 30; Foreman, supra note 31.

33 . See, e.g., State v. Lohmeier, 205 Wis.2d 183, 556 N.W.2d 90, 93 (1996) (in vehicular homicide case where defendant argued that jury instructions violated due process because they denied him a "meaningful opportunity” for the jury to consider his defense, the court held there was no reasonable likelihood that interplay of legally correct affirmative defense and contributory negligence instructions misled jury into applying reasonable doubt standard in an unconstitutional manner).

34 . See, e.g., State v. Burris, No.2009AP956CR, 2010 WL 273952, at *6, 8 (Wis.App.Jan.26, 2010), cert. granted, 329 Wis.2d 371, 791 N.W.2d 380 (2010) (in reviewing convictions for reckless bodily injury while armed and firearm offense, appellate court reversed because of "a reasonable likelihood that the jury was misled” by a "legally accurate” response to the jury’s questions "and therefore applied potentially confusing instructions in an unconstitutional manner”) (quoting Lohmeier, supra note 33, 556 N.W.2d at 94).

35 . Kotteakos v. United States, 328 U.S. 750, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946); see, e.g., Shelton v. United States, 983 A.2d 979, 985 (D.C.2009) (affirming conviction for unlawful possession of controlled substance because trial court’s erroneous instruction that conviction required no mandatory period of incarceration was harmless based on strength of government's case); Yelverton, supra note 23, 904 A.2d at 389.

36 . Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967); see, e.g., Johnson v. United States, 980 A.2d 1174, 1181 (D.C.2009) (affirming conviction for second-degree murder while armed, despite erroneous aiding-and-abetting instruction, because error was harmless beyond a reasonable doubt given strength of government’s case that left "no reasonable possibility” that the jury "relied on the crutch of the erroneous instruction”); Tyree v. United States, 942 A.2d 629, 638-40 (D.C.2008) (affirming first-degree murder conviction, despite erroneous aiding- and-abetting instruction that eliminated mens rea requirement, because court concluded beyond a reasonable doubt that any impartial trier of fact who credited the government’s evidence would be bound to conclude that defendant intended to kill the decedent).

37 . Arizona v. Fulminante, 499 U.S. 279, 307-08, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991).

38 . Id. at 309, 111 S.Ct. 1246.

39 . Id. at 310, 111 S.Ct. 1246.

40 . 508 U.S. 275, 282, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993). In Sullivan, the Court observed that "the Fifth Amendment requirement of proof beyond a reasonable doubt,” inherent in due process, and "the Sixth Amendment requirement of a jury verdict are interrelated.” Id. at 278, 113 S.Ct. 2078. The Sixth Amendment would not be satisfied unless the jury, properly instructed, were to find guilt under the reasonable doubt standard. It follows that a finding of guilt under an erroneous reasonable doubt instruction cannot amount to a "jury verdict within the meaning of the Sixth Amendment,” and thus that kind of error cannot possibly be harmless. Id. at 280, 113 S.Ct. 2078. The instructional error, therefore, was "structural,” requiring automatic reversal. Id. at 282, 113 S.Ct. 2078 (citing Fulminante, supra note 37, 499 U.S. at 309, 111 S.Ct. 1246).

41 . Id. at 281, 113 S.Ct. 2078 (emphasis in original).

42 .See id. at 281-82, 113 S.Ct. 2078.

43 . Davis, supra note 23, 510 A.2d at 1053 (quoting Arroyo v. Jones, 685 F.2d 35, 39 (2d Cir.), cert. denied, 459 U.S. 1048, 103 S.Ct. 468, 74 L.Ed.2d 617 (1982)).

44 . Id. (citations omitted).

45 . Smith, supra note 9, 709 A.2d at 83 (citation omitted). Four years earlier in Butler, supra note 31, 646 A.2d at 337, this court had warned with respect to the previous Redbook instruction: "We have approved a standard [reasonable doubt] instruction; we are comfortable with it; and a trial court takes unnecessary risks with the criminal justice process by tinkering with it. We therefore wish to make clear that, given the critical significance of the reasonable doubt instruction, for purposes of analysis in the future we shall consider trial court deviations from that instruction, over defense objection, to be improper — indeed, presumptively erroneous.” (Emphasis added.)

46 .Smith, supra note 9, 709 A.2d at 80. We also worried that variations in the reasonable doubt instruction would create the appearance, if not the reality, of unfairness in our justice system; absent uniformity, a defendant convicted under one formulation of the instruction might well wonder what would have happened had he been tried in another courtroom. Finally, without a mandated instruction, the situation would persist in which deviations from the traditional "Redbook instruction” would foster "repetitive constitutional challenges” that burden the court system. Id. at 81.

47 . Supra note 31, 633 A.2d at 794 (internal quotation marks omitted) (finding no plain error when trial judge, in defining reasonable doubt, substituted the phrase "a deep rooted belief in” defendant’s guilt for "an abiding conviction of” guilt).

48 . Butler, supra note 31, 646 A.2d at 337 & n. 7.

49 . Proctor, supra note 31, 685 A.2d at 740 (citing Foreman, supra note 31, and quoting Butler, supra note 31).

50 .In State v. Dreher, 302 N.J.Super. 408, 695 A.2d 672 (1997), overruled on other grounds by State v. Brown, 170 N.J. 138, 784 A.2d 1244 (2001), a murder and conspiracy case, the court rejected appellant's contention that "the judge's supplemental instruction on reasonable doubt, which diverged from the model jury charge, diluted the State’s burden of proof.” 695 A.2d at 700. The court initially instructed with the model charge; then, when the jury asked it to "redefine reasonable doubt,” id., the court reinstructed by adding to the model charge three sentences, including one from Webster's dictionary. Appellant argued that "the supplemental charge was unbalanced in the State’s favor” because "there was only one sentence explaining what reasonable doubt is but three or four sentences explaining what it is not.” Id. "These errors were compounded,” appellant argued, "because they were set forth in response to a jury question and in a supplemental charge,” which, according to appellant, is "given greater attention by a jury.” Id. at 700-01. The court rejected appellant's argument because, although the supplemental charge added language to the model charge "in two minor respects,” when taken "as a whole the judge's charge did nothing to lessen the State’s burden of proving defendant guilty beyond a reasonable doubt.” Id. at 702.

⁵¹

52 . In Payne, after instructing that "the government never has to prove guilt beyond all doubt,” the court added, "That’s impossible ” (emphasis added). Then came: "They do not have to prove guilty beyond a shadow of a doubt. There’s no such thing" (emphasis added). See text accompanying supra note 13. Counsel for co-defendant Carter asked for omission of the italicized sentences as "empty emphasis” and "cheerleading”; the government did not object; and the judge agreed, observing that "those two phrases perhaps overdo it a bit.” Rather than adding "empty emphasis,” however, these two omitted sentences arguably helped avoid creating a three-tier structure of doubt, with reasonable doubt on the bottom (which may be why they prompted no objection from trial counsel in Payne). That is, by emphasizing the nonexistence of the kind of "doubt” each referred to, these sentences arguably helped create the impression that both concepts were identical and thus not, as appellant has argued, two, separate levels of doubt above reasonable doubt. The government has not pressed such an argument, however; in fact, on appeal the government has argued that removal of these ‘‘most argumentative phrases” from the Payne reinstruction removed the kind of emphasis that counsel was complaining about.

53 . See supra Part II. B.

54 . Appellant acknowledges, see supra Part II. B., that some jurors may have equated the new formulation, "beyond a shadow of a doubt,” with "beyond all doubt” as used in the initial, Smith instruction, whereas other jurors may have perceived a new, three-level structure of doubt. But, he adds — and we agree — however the jurors' perceptions were allocated, die addition of this "shadow” language to the reinstruction created ambiguity. See Estelle, supra note 30, 502 U.S. at 72, 112 S.Ct. 475; Foreman, supra notes 31 and 47, 633 A.2d at 794. Under all the circumstances, we conclude, as elaborated more fully below, that this ambiguity, when coupled with the addition of more emphatic, repetitive language stating what reasonable doubt is "not,” created a reasonable likelihood that the jury — which had deliberated for more than four days but returned a verdict of guilty within two hours of reinstruction — "applied the instructions in a way that violated the Constitution” by unacceptably lowering the standard of reasonable doubt. Victor, supra note 30, 511 U.S. at 22-23, 114 S.Ct. 1239; Foreman, supra notes 31 and 47, 633 A.2d at 794.

55 .Appellant's argument receives support from a recent Wisconsin case, State v. Burris, supra note 34, 2010 WL 273952. The defendant was charged with recklessly causing bodily injury by shooting the son of his ex-girlfriend in the neck. In the instruction informing the jury that it must consider whether the "conduct showed utter disregard for human life,” the court added that the jury also could consider whether the conduct "showed any regard for human life.” Id. at *4. Thereafter, the jury sent the judge a note asking whether it "could consider facts and circumstances after the shooting.” Id. at *6. Instead of simply answering "yes,” which both the state and the defendant agreed would be the correct answer, the judge read language to the jury from another case discussing ‘‘[a]fter-the-fact regard for human life” in the context of evaluating whether the defendant had shown "utter disregard” for life. Id. Both parties, as well as the reviewing court, agreed that the reinstruction was a correct statement of the law. In the appellate court's judgment, however, the language of reinstruction confused the issue and misled the jury into believing that after-the-fact behavior was irrelevant. The Wisconsin court accepted the defendant's argument:

[T]he point is not that the trial court’s answer was an incorrect statement of the law in the abstract; rather, the point is that — in the specific context in which it was provided, as a response to the jury’s question about whether it could consider after-the-fact conduct — the answer was misleading because it implied that the jury should not consider after-the-fact regard at all, or at least not consider it equally with other circumstances.

Id. at *7 (emphasis in original). The court therefore reversed, concluding that “there is a reasonable likelihood that the jury was misled and therefore applied potentially confusing instructions in an unconstitutional manner.” Id. (quoting Lohmeier, supra note 33, 556 N.W.2d at 90) (internal quotation marks omitted) (relying inter alia, id. at 93, on Estelle and Victor, supra note 30).

56 . Compare Winters v. United States, 317 A.2d 530 (D.C.1974) (en banc) (approving modified form of anti-deadlock charge for future cases) with Epperson v. United States, 495 A.2d 1170, 1176 (D.C.1985) (after giving anti-deadlock instruction to "hung jury,” trial court erred in giving second anti-deadlock instruction because two such instructions called “into question the integrity of the verdict” and thus warranted reversal).

57 . See text supra at note 11.

58 . See id.

⁵⁹

60 . See text supra at pages 770-71.

⁶¹

62 . Smith, supra note 9, 709 A.2d at 79.

63 . See id. at 83.

64 . Id. (citation and internal quotation marks omitted).

65 . To be clear: in stressing that the trial court should not stray from Smith while instructing or reinstructing on "reasonable doubt,” we have never said, and we do not say now, that a failure to use Smith's language automatically violates due process. See, e.g., Brown v. United States, 881 A.2d 586, 596 (D.C.2005) (trial court's mistaken use of pre-Smith Redbook instruction on "reasonable doubt" was not, after review, "constitutionally deficient”); Butler, supra notes 31 and 45, 646 A.2d at 337 (omission of clause required by standard Redbook instruction on "reasonable doubt,” while "improper,” did "not create a constitutionally deficient instruction under the particular circumstances of this case”). Accordingly, our en banc rule making the Smith instruction mandatory reflects an effort to avoid constitutional error, not to create such error per se upon violation.

66 . Davis, supra note 23, 510 A.2d at 1053.

67 . See supra note 56 and accompanying text.

68 . In the words of the U.S. Court of Appeals for the Fifth Circuit:

In his charge to the jury as well as in his examination of witnesses, the trial judge must be extremely careful to refrain from becoming an advocate for the Government .... It is well known ... that juries are highly sensitive to every utterance by the trial judge, the trial arbiter, and that some comments may be so highly prejudicial that even a strong admonition by the judge to the jury, that they are not abound by the judge's views, will not cure the error.

United States v. Musgrave, 444 F.2d 755, 761, 762 (5th Cir.1971) (reversible error because trial judge, in charge to jury, made prejudicial comments in government's favor without further instruction that jury was not bound by judge's comments or questions) (emphasis added). See also 1 Weinstein’s Federal Evidence § 107.23[5][b] (2d ed.2011) (“In criminal cases, the trial judge must never tell the jury that he or she thinks the defendant is guilty, either expressly or by implication.”) (emphasis added); Walter W. Jones, Annotation, Propriety and Prejudicial Effect of Federal Judge’s Expressing to Jury His Opinion as to Defendant’s Guilt in Criminal Case, 7 A.L.R. Fed. 377 § 2 (1971) (“It is generally recognized that in the absence of exceptional circumstances, remarks made by a trial judge to the effect that he believes the accused to be guilty of the crime with which he is charged are improper, prejudicial, and incapable of being cured by other instructions or remarks. Such remarks have been held especially improper and especially prejudicial where the judge has made them after the jury has indicated its inability to agree on a verdict.”) (footnotes omitted) (emphasis added).

69 . Foreman, supra note 31, 633 A.2d at 794 (quoting Merlos, supra note 31, 299 U.S.App. D.C. at 403, 984 F.2d at 1241 and Estelle, supra note 30, 502 U.S. at 72, 112 S.Ct. 475) (internal quotation marks omitted); (accord, Victor, supra note 30, 511 U.S. at 22-23, 114 S.Ct. 1239).

70 . Sullivan, supra note 40, 508 U.S. at 281, 282, 113 S.Ct. 2078.

71 . Victor, supra note 30, 511 U.S. at 22-23, 114 S.Ct. 1239.

72 . 527 U.S. 1, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999).

73 .We respond to our concurring colleague's footnote 6. The judge answered the jury's question about the meaning of reasonable doubt. Our decision addresses that issue— the issue raised by appellant. If, as a result of reinstruction, one or more jurors inferred that the judge believed appellant was guilty, then we believe that this inference would have been derived, in context, from a perception that the judge believed his original instruction had set the bar for reasonable doubt too high, and that "some change” — some ratcheting downward in the government's favor — was necessary to achieve the correct verdict. Our colleague's theory builds on top of our analysis, creating an a fortiori case; it does not swallow and eliminate the majority analysis.