It is a salutary and perhaps even common practice for trial judges in criminal cases, when appropriate, to instruct jurors to suppress any temptation they may have to visit the location of the alleged crime on their own. The trial judge regretted not having done so in this case, for, as it happened, a juror ventured out to the scene of the alleged drug sale the night before deliberations began to examine the lighting conditions. Appellant’s defense was mis-identification, and, as the trial judge later acknowledged, the lighting was “important, even central” to at least one factor bearing on the reliability of the police officer’s identification. Moreover, when the jury foreperson brought the matter to the judge’s attention soon after deliberations began, the juror in question confessed that as a result of the visit, “I kind of drew my own conclusion as to what I felt the lighting condition was.” Nonetheless, responding to a question by the judge, the juror added that he would have “no problem” deciding the case solely upon the evidence heard in court.
The government argues that, under the deferential standard of review appropriate in these circumstances, we must sustain the trial judge’s acceptance of the juror’s assurance and refusal to grant a mistrial. We conclude, to the contrary, that in the circumstances of this single-witness identification case, even conceding the trial judge’s superior vantage point from which to evaluate prejudice, the risk of a tainted verdict from the juror’s conduct is too great to permit the judgment to stand. We therefore reverse.
I.
The government’s evidence showed that Officer Gary Curtis, dressed in plainclothes, approached Rhonda Tinch in the 1200 block of Wiley Street, N.E. at about 1:55 a.m., and asked her for a “twenty” of crack cocaine. Tinch told him to follow her to the middle of the block where she approached appellant, spoke with him, and received from him a plastic baggie which she gave to Officer Curtis in return for twenty dollars in pre-recorded police funds. The trio were about arm’s length apart during the transaction. Tinch gave the money to appellant, who in turn gave it to a fourth individual, Adams. Curtis returned to his vehicle and radioed a description of the three sellers.1 Appellant and the others were then arrested on the scene and identified by Curtis; the twenty dollars in police funds was found on Adams.2 Cur*682tis testified that the scene was illuminated by high intensity street lamps and that, in keeping with his experience, he had focused on the features of the sellers so he could later identify them. His encounter with appellant lasted “about 40 seconds.” One defense witness, who claimed to be on the scene with appellant at the time but denied seeing him involved in a drug sale, confirmed that the area “was light.” But a defense investigator who had inspected the scene testified in detail about shadows cast by the street lights on this particular block. Defense witnesses also testified that other black males matching the description by Curtis were on the scene at the time, though arrest team officers insisted there were only two such males and that neither matched Curtis’s description of appellant.
II.
Were we reviewing the legal sufficiency of the evidence, we would not hesitate in' concluding that reasonable jurors could fairly credit Officer Curtis’s identification of appellant. Nevertheless, issue was joined on the question of identity and specifically on Curtis’s ability to identify appellant — within the space of some 40 seconds — given the lighting conditions on the scene, and it is in this context that appellant’s claim of prejudicial misconduct by a juror comes before us. The issue arose shortly after the trial judge instructed the jury and dismissed the alternate jurors, and the panel retired to deliberate. The judge received a note from the foreperson stating that a juror had visited the crime scene, but that as soon as he revealed that fact, deliberations had stopped and he had said nothing further to the other jurors. Although appellant immediately requested a mistrial, the judge instead — quite correctly — ordered the juror brought into the courtroom where he questioned him as follows:
We understand that you went to the scene last night for the purpose of making your own independent investigation; is that right?
JUROR NO. 737: Yes.
THE COURT: What — where is it that you went precisely?
JUROR NO. 737: I just drove through Wiley Street. I didn’t stop anywhere in particular, I just drove through.
THE COURT: Okay.
What was it that you were curious to see?
JUROR NO. 737: The lighting conditions.
THE COURT: Do you feel that having seen what you saw gives you a different perspective than you had at the close of the evidence yesterday before you went to Wiley Street last night?
JUROR NO. 737: Well, no, sir.
THE COURT: Well, but let me ask it a different way. You had heard the testimony, and I think there was not much more today about lighting. Most of the testimony about the lighting — I guess the real — the issue sort of came to a head with the conclusion of the defense investigator’s testimony yesterday about whether or not — about whether there were shadows cast by the street lamps and what not.
After you drove through the street, did you have a better understanding or different understanding — I don’t want to say better, I don’t want to put a value on it — did you have a different understanding of the evidence than you had when you left here at five o’clock yesterday?
JUROR NO. 737: Yes, sir, I think I did.
THE COURT: In what way?
JUROR NO. 737: I kind of drew my own conclusion as to what I felt the lighting condition was.
THE COURT: Counsel, approach.
By the way, what time was it approximately when you drove through?
JUROR NO. 737: Eleven o’clock p.m.
At the parties’ request, the judge asked further questions and the juror explained that, when deliberations had begun and it became his turn to speak, “I stood up and opened up the chart and ... said, I went to the area last night,” whereupon the fore*683person “stopped me right there” and reported the disclosure to the court. The judge continued the inquiry:
Tell me this, would it be possible for you — and I really need your best answer on this — without any regard to what my decision in this thing is going to be, do you think it would be possible for you to go back in the jury room and deliberate with the other jurors, the evidence — see the case, the trick is this, the case has to be decided on the evidence that you hear in court.
JUROR NO. 737: Uh-huh.
THE COURT: That’s our notion of how it’s fair.
Do you think that you could now decide the case on the evidence you heard in court, or is it — is that likely humanly impossible that you would have to be influenced by what you saw yourself?
JUROR NO. 737: No, sir, I would have no problem with that.
THE COURT: Okay. All right. Thank you very much for your honest answers.
After hearing argument from defense counsel, the judge reviewed the factors bearing on the issue of a juror’s contact with extra-judicial information and when such contact requires a mistrial. The judge then made findings, beginning with the fact that “the taint is limited to one juror” inasmuch as the juror in question had “not share[d] ... the result of his investigation with the other jurors.”3 Further, while finding the “possible prejudice” to the defense (and “equally [to the government]”) from the juror’s conduct to be “great,” the judge was “persuaded by the juror’s seeming candor in his statement that he could decide this case based on the evidence he heard in court without regard to his own independent investigation.” Moreover, while agreeing that “the lighting conditions on the street at nighttime” were “important, even central to at least one of the facts to consider in weighing the identification testimony,” the judge pointed out that he had instructed the jury on other factors as well bearing on the reliability of the identification, “such as the length of the encounter, the description given, [and] the certainty or lack thereof of the subsequent identification.” Lastly, adverting to the testimony of defense witnesses about the lighting, the judge found that this “is not a case where there just were prosecution witnesses, police officers, and no defense other than the cross examination.” He therefore ruled:
[C]onsidering the total circumstances ext[a]nt in this case, I believe that — and the juror’s response that he can decide this case and continue in deliberations without regard to what he saw last night, I believe that these factors weight against declaring a mistrial in this case. And I further believe that the outcome of this trial will not be tainted by the juror’s indiscretion in visiting the scene.
III.
In a recent case where analogous juror partiality was claimed (the juror, originally an alternate, had improperly discussed the case with another alternate after both had initially been dismissed), we stated that “the remedy for allegations of juror bias is a hearing to determine whether the misconduct actually resulted in prejudice.” Leeper v. United States, 579 A.2d 695, 698 (D.C.1990) (quoting Washington v. Washington Hospital Center, 579 A.2d 177, 185 (D.C.1990)). See also Harris v. United States, 606 A.2d 763, 765 (D.C.1992). The trial judge conducted such an inquiry in this case, and we find no deficiency in it. Indeed, appellant’s primary complaint is not with the hearing conducted, but with its outcome. Here too, however, appellant’s burden is not light, for we have held that,
[fallowing a proper hearing, the determination of juror bias or prejudice lies particularly within the discretion of the trial court, reversible only for a clear abuse of discretion, and the findings of fact under*684lying that determination are entitled to “great deference.”
Leeper, 579 A.2d at 698 (citations omitted). “Our review is deferential because the question of prejudice turns substantially on the judge’s appraisal of the juror’s demean- or, and is therefore one about which the trial judge is 'especially qualified to render a sound opinion.’ ” Id. (citations omitted).
In this case, the trial judge was impressed with the juror’s “seeming candor” in stating that he could put aside the results of “his own independent examination.” Certainly the juror was candid in acknowledging — without knowing, perhaps, the legal implications of his admission— that from his visit to the scene he “kind of drew [his] own conclusion as to what I felt the lighting condition was.” We are bound to give deference, even great deference, to the judge’s appraisal of the juror’s assurance that he could set this conclusion aside and reach a decision only upon the evidence heard in court.4 Even so, it is the government’s burden to demonstrate that the juror’s contact with extraneous information was harmless or non-prejudicial. E.g., United States v. Butler, 262 U.S.App.D.C. 129, 134, 822 F.2d 1191, 1196 (1987). See Remmer v. United States, 347 U.S. 227, 229, 74 S.Ct. 450, 451, 98 L.Ed. 654 (1954) (stressing government’s burden “to establish ... that such contact with the juror was harmless to the defendant”). This point requires elaboration. In its brief the government does not appear to dispute placement of the ultimate burden of showing harmlessness on the prosecution. It does resist strongly appellant’s reliance on Remmer, supra, for the principle that juror partiality is “presumed” or “imputed” from the fact alone of the juror’s exposure to extraneous material. In this the government is correct. Following the Supreme Court’s decision in Smith v. Phillips, 455 U.S. 209, 102 S.Ct. 940, 71 L.Ed.2d 78 (1982), this court has made clear its understanding that the remedy for a claim of juror partiality or taint is a thorough inquiry by the trial judge into whether the defendant suffered actual prejudice. Harris, 606 A.2d at 765; Artisst v. United States, 554 A.2d 327, 330 n. 3 (D.C.1989). Indeed, in Harris we recently iterated that “[w]here a claim of juror partiality is made, ‘the remedy ... is a hearing in which the defendant has the opportunity to prove actual bias_” 606 A.2d at 765 (emphasis added) (quoting Shannon & Luchs Management Co. v. Roberts, 447 A.2d 37, 41 (D.C.1982) (in turn quoting Phillips, 455 U.S. at 215, 102 S.Ct. at 945)). But where, following a hearing, the defendant has established a substantial likelihood of actual prejudice from the unauthorized contact, we agree with the United States Court of Appeals for the District of Columbia Circuit that “all reasonable doubts [about the juror’s ability to render an impartial verdict must] be resolved in favor of the accused.” United States v. Williams, 262 U.S.App.D.C. 112, 128, 822 F.2d 1174, 1190 (1987). In this sense “Remmer’s allocation of the burden [of proving harmlessness to the government] remains the law,” Butler, 262 U.S.App.D.C. at 133-34 n. 2, 822 F.2d at 1195-96 n. 2.5
On the facts of this case, even conceding the trial judge’s superior vantage point from which to assess prejudice, we are not persuaded that the government has met the burden the law imposes on it in this context. Basic to our analysis are the circumstances surrounding the juror’s conduct and the effect it had upon his under*685standing.6 First, courts have been unwilling to ignore the real possibility of prejudice where the improper contact furnished the juror with “crucial extra-judicial information,” Butler, 262 U.S.App.D.C. at 134, 822 F.2d at 1196, or information “concerning] the primary factual issue before the jury.” United States v. Delaney, 732 F.2d 639, 642 (8th Cir.1984). Specifically, whether a conviction may stand despite an unauthorized visit to the scene has often turned upon whether, for example, the physical location or condition of the crime scene — including the lighting — was disputed or whether the juror(s) saw only what the trial evidence concededly had reflected. See Annotation, Unauthorized View of Premises by Juror or Jury in Criminal Case as Ground for Reversal, New Trial, or Mistrial, 50 A.L.R. 4th 995, 1017-27 (1986). The government admits that mis-identification was the heart of appellant’s defense, but points to the trial judge’s comment that the lighting conditions were only “one of the factors” bearing on the reliability of Officer Curtis’s identification. Yet, as the trial judge found, the varying testimony about lighting made that issue “important, even central” to the adequacy of the officer’s opportunity to observe appellant. Moreover, the lighting was undeniably an important issue for this juror, who purposely “went to the area” the night before deliberations to inspect those conditions himself.
Second, the situation before us is unlike the more typical one where, despite exposure to extra-judicial matter, the juror is able to represent that this had no effect on his understanding of, or judgment about, the case. E.g., Butler, supra, 262 U.S.App.D.C. at 134-35, 822 F.2d at 1196-97. After visiting the scene, the juror here “kind of drew [his] own conclusion” about the lighting. Whether he might have reached that same conclusion based on the trial evidence — indeed, whether we think that evidence left room for a reasonable doubt about the adequacy of the lighting— is not the point; this juror felt uncertain enough about the issue to inspect the conditions for himself and reach his “own conclusion” from that visit. Although he assured the judge he would still have “no problem” deciding the case on the evidence presented, the mental somersaults — in appellant’s phrase, the “optical compartmentalization” — this required must induce considerable skepticism. Cf. Bruton v. United States, 391 U.S. 123, 135-36, 88 S.Ct. 1620, 1627-28, 20 L.Ed.2d 476 (1968). The principle that a juror “ ‘is well qualified to say whether he has an unbiased mind in a certain matter,’ ” Phillips, 455 U.S. at 217 n. 7, 102 S.Ct. at 946 n. 7 (quoting Dennis v. United States, 339 U.S. 162, 171, 70 S.Ct. 519, 523, 94 L.Ed. 734 (1950)), loses much of its pertinency when that same juror has declared himself of two minds— one saying he has reached his own conclusion from the extraneous contact, the other saying he can lay that conclusion aside.7 As we stated earlier, “all reasonable doubts” about whether this sort of compartmentalization was possible in a case where the lighting conditions were important and disputed must “be resolved in favor of the accused.” Williams, 262 U.S.App.D.C. at 128, 822 F.2d at 1190.
Finally, we cannot disregard the fact that, before the trial judge was able to intervene at all, the juror had told the other members of the jury of his visit to the *686scene, and then returned to the deliberations after being questioned by the judge. The government points out that appellant requested no relief (short of a mistrial) for this disclosure such as a voir dire of the jury or, at a minimum, an instruction to the juror in question not to tell the others of his “conclusion.” It is true that defense counsel shares an obligation with the judge to fashion the proper inquiry in these circumstances, for “[wjhat Remmer [, supra,] demands is notice to the accused of any juror contact, and an opportunity for the accused to participate in any proceeding to determine its impact.” Williams, 262 U.S.App.D.C. at 128, 822 F.2d at 1190 (emphasis added); Leeper, 579 A.2d at 700. Nonetheless, even if the fact of disclosure to the other jurors would not itself justify reversal, it is a factor that must be given weight in our assessment of prejudice.
All told, despite the conscientious trial judge’s effort to cope with the intrusion of extraneous matter into the deliberations, we are left with too great an uncertainty about the effect of the juror’s conduct on his — and the jury’s — ability to render an impartial verdict to permit the judgment to stand.
Reversed and Remanded for a New Trial.
. He described appellant as a Black male wearing a blue jacket and blue jeans.
. The trial judge gave a missing evidence instruction in view of Curtis’s failure to preserve the piece of paper on which he claimed to have *682recorded the serial number of the twenty dollar bill.
. The judge stated:
It appears as if the other jurors realized the impropriety of what this juror had done and seized [sic; ceased] their deliberations and made the problem known to the Court.
. Appellant suggests that it is not entirely clear what the juror meant by his terse response of "no problem” to the judge’s somewhat elaborate question eliciting that response. But appellant did not ask the court to clarify by additional questions whether the juror (a) understood his obligation to decide the case solely on the evidence and (b) was insisting that he could do so. The judge found the juror had stated that he could put aside the "conclusion” of his independent investigation, and we accept that finding as to what the juror asserted he could do.
. The Butler court noted that no "federal appellate court [other than the Sixth Circuit had] departed from Remmer's statement of the legal standard for evaluating the effect of an improper judicial contact." 262 U.S.App.D.C. at 133 n. 2, 822 F.2d at 1195 n. 2.
. As an original matter, it is not easy to discount the psychological fact that a juror told he has engaged in unauthorized conduct may seek to make amends by promising objectivity he can no longer exercise. This court, however, in light of Phillips, supra, has rejected the notion that a juror's testimony in insisting he or she can remain impartial despite an extra-judicial contact "is inherently suspect." Catlett v. United States, 545 A.2d 1202, 1213 n. 22 (D.C.1988) (citing Phillips, 455 U.S. at 217 n. 7, 102 S.Ct. at 946 n. 7), cert. denied, 488 U.S. 1017, 109 S.Ct. 814, 102 L.Ed.2d 803 (1989).
. See Ex parte Lasley, 505 So.2d 1263, 1264 (Ala.1987) (application of rule requiring reversal for juror misconduct that might have influenced the verdict "cannot in all cases depend entirely upon the jurors’ statements that the extraneous information did not affect their verdict," for “[t]he jurors cannot in every case determine the question of whether they were, or might have been, improperly influenced").