This juvenile delinquency case is more than five years old. On January 20, 1999, following a factfinding hearing, As.H., then sixteen years of age,1 was adjudicated guilty of robbery. The sole evidence implicating As.H. in the offense was the testimony of the victim, Ms. Michal Freedhoff, who identified As.H. at a photo array almost a month after the robbery and again in court more than four months after that. Ms. Freedhoff described her level of certainty on both occasions, however, as “seven or eight” on a scale of one to ten. Because Ms. Freedhoff was obviously less than positive regarding her identification, and for other reasons described below, we conclude as a matter of law that the evidence was insufficient to prove beyond a reasonable doubt that As.H. was involved in the robbery. Accordingly, we reverse.
I.
In the early morning hours of August 17, 1998, between 12:30 and 1:00 a.m., Ms. Freedhoff was robbed by three2 or more *458young men. The assailants knocked Ms. Freedhoff to the ground, threatened her with “a long piece of wood” which, Ms. Freedhoff believed, was “suppose[d] to look like a rifle,’’ordered her to “shut up, bitch,” and robbed her of her purse and her personal electronic organizer. Ms. Freedhoff promptly reported the crime to the police. Officers detained a group of young men shortly after the robbery and arranged a show-up, but Ms. Freedhoff stated that the detained individuals were not the robbers. Indeed, she was “completely” certain that the individuals at the show-up were not the guilty parties.
Ms. Freedhoff testified that there were street lights in the area where the robbery occurred. She further stated that she had been outside in the street for some time, so that her eyes had become accustomed to the dark. Nevertheless, Ms. Freedhoff could not provide an informative description of her assailants. According to Detective Ross, she recalled nothing distinctive about their clothing; “young black males and baggy clothes” was his recollection of her report. At the factfinding hearing, which took place more than five months after the robbery, Ms. Freedhoff recalled that the robbers were teenagers, “two dark-skinned and one light,” each of a different height, and that “one had shorts and sneakers and another may have had a hat.” Ms. Freedhoff was also uncertain as to the role which the individual she tentatively identified as As.H. allegedly played in the robbery.
On September 11, 1998, Detective Ross showed Ms. Freedhoff an array of nine Polaroid pictures and asked her if she recognized anyone who was involved in the offense. At a hearing on As.H.’s motion to suppress identification, Ms. Freedhoff testified as follows regarding this array:
Q: Now, Ms. Freedhoff, on that day did you identify any of the people in the photos as having been involved in the incident of August 16th?
A: Yes, I did.
Q: Which photos did you identify?
A: These two marked nine and [ten] I was very certain about and the two marked three and four I was less certain about.
Q: During the identification procedure, did you talk to the detective about your level of certainty?
A: Yes.
Q: In terms of nine and [ten], what was your level of certainty that those people were involved?
A: I [was] asked to rate them on a scale of I believe it was one to [ten] and I believe I said it was, that nine and [ten], I was seven or eight.
Q: And in terms of three and four, how did you rate those?
A: Six.
According to Ms. Freedhoff, the photograph of As.H. was No. 10. At the fact-finding hearing, Ms. Freedhoff initially stated that she saw one of the robbers sitting in the courtroom, pointing out As.H. When asked which of the individuals in the array he was, Ms. Freedhoff “believed” that it “would be Number 10.” However, when counsel for the District of Columbia again asked Ms. Freedhoff about her present level of certainty in making the identification — how certain are you?— the witness adhered to her previous estimate: “At the time, on a scale of one to [ten], I said that I was .seven or eight.”
According to Detective Ross, who also testified regarding the viewing of the photo array, Ms. Freedhoff was “comfortable in saying they could be the people that *459robbed her.”3 (Emphasis added.) Ross further disclosed that he “may have discussed with [Ms. Freedhoff] that I had a previous history with the persons that she had picked. They were my possible suspects in the case.”
Without elaborating on his reasons, the trial judge denied As.H.’s motion to suppress identification4 and found As.H. guilty as charged. This appeal followed.
II.
In evaluating claims of evidentiary insufficiency in juvenile delinquency appeals, we view the record “in the light most favorable to the [District], giving full play to the right of the judge, as the trier of fact, to determine credibility, weigh the evidence, and draw reasonable inferences .... We will reverse on insufficiency grounds only when the [District] has failed to produce evidence upon which a reasonable mind might fairly find guilt beyond a reasonable doubt.” In re T.M., 577 A.2d 1149, 1151 (D.C.1990) (citations omitted); see generally Rivas v. United States, 783 A.2d 125, 133-35 (D.C.2001) (en banc). “Even identification testimony of a single eyewitness will be sufficient so long as a reasonable person could find the identification convincing beyond a reasonable doubt.” Peterson v. United States, 657 A.2d 756, 760 (D.C.1995) (citations and internal quotation marks omitted). Moreover, the District was not required to prove As.H.’s guilt beyond all doubt. “There is no rule of law which requires an identification to be positive beyond any shadow of doubt.” People v. Spinello, 303 N.Y. 193, 101 N.E.2d 457, 462 (1951) (citations omitted).
Nevertheless, the “reasonable doubt” standard of proof is a formidable one. It “requires the factfinder to reach a subjective state of near certitude of the guilt of the accused.” Rivas, 783 A.2d at 133 (quoting Jackson v. Virginia, 443 U.S. 307, 315, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979)). Although appellate review is deferential, we have “the obligation to take seriously the requirement that the evidence in a criminal prosecution5 must be strong enough that a jury6 behaving rationally really could find it persuasive beyond a reasonable doubt.” Id. at 134. Moreover, “while [the trier of fact] is entitled to draw a vast range of reasonable inferences from evidence, [he or she] may not base [an adjudication of guilt] on mere speculation.” Id. (citation omitted).
In the present case, we have an eyewitness identification of questionable certitude, and the witness and the respondent are strangers. Ms. Freedhoff saw her assailants at night and under extremely stressful conditions. Moreover, this is a “pure” eyewitness identification case; there is no evidence linking As.H. to the robbery except for Ms. Freedhoffs statements upon viewing the array almost a month after the event and her testimony at the factfinding hearing more than five months after she was robbed.
The vagaries of eyewitness identification, and the potential for wrongful convictions or adjudications based upon such evidence, have long been recognized in the *460District of Columbia. United States v. Telfaire, 152 U.S.App. D.C. 146, 149-51, 469 F.2d 552, 555-57 (1972) (per curiam); Crawley v. United States, 320 A.2d 309, 311-12 (D.C.1974).7 More recently, in Webster v. United States, 623 A.2d 1198 (D.C.1993), we summarized this concern as follows:
“[T]he identification of strangers is proverbially untrustworthy. The hazards of such testimony are established by a formidable number of instances in the records of English and American trials.” FELIX FRANKFURTER, THE CASE OF SACCO AND VANZETTI (1927), quoted in United States v. Wade, 388 U.S. 218, 228, 87 S.Ct. 1926, 18 L.Ed.2d 1149 (1967). Indeed, “[positive identification of a person not previously known to the witness is perhaps the most fearful testimony known to the law of evidence.” Wehrle v. Brooks, 269 F.Supp. 785, 792 (W.D.N.C.1966), aff'd, 379 F.2d 288 (4th Cir.1967). Even if the witness professes certainty, “it is well recognized that the most positive eyewitness is not necessarily the most reliable.” Crawley ..., 320 A.2d [at] 312.
Id. at 1204 n. 15. Here, the witness did not even profess certainty. Moreover, the present case concerns a hesitant cross-racial identification by a white woman of a black teenager, and “[i]t is well established that there exists a comparative difficulty in recognizing individual members of a race different from one’s own.” ELIZABETH LOFTUS, EYEWITNESS TESTIMONY § 4.9 (3d ed. 1997 & Supp.1999); see State v. Cromedy, 158 N.J. 112, 727 A.2d 457, 461-68 (1999) (discussing at length the difficulties in cross-racial identification and mandating a jury instruction on the subject in some cases); John P. Rutledge, They All Look Alike: The Inaccuracy of Cross-Racial Identifications, 28 AM. J. CRIM. L. 207 (2001).
It is in the context of these realities that we now turn to the dispositive issue in this appeal, namely, whether Ms. Freedhoffs testimony — the only evidence of As.H.’s participation in the robbery — was legally sufficient to support a finding of guilt beyond a reasonablé doubt. The key fact is that, both when viewing the polaroid photographs and when testifying in open court, Ms. Freedhoff candidly characterized her level of “certainty” — ie., of her being “very certain” — as seven or eight on a scale of one to ten. Her testimony leads inexorably to the conclusion that her level of un certainty — ie., the possibility that As.H. was not involved — was two or three out of ten — a 20% to 30% possibility of innocence. This differs dramatically from Ms. Freedhoffs complete certainty that the young men she viewed at the show-up on the night of the offense were not the robbers. The contrast between Ms. Freedhoffs statements in the two situations is revealing, and surely negates the “near certitude,” Rivas, 783 A.2d at 133, that is required for a showing of guilt beyond a reasonable doubt. The “seven or eight out óf ten” assessment is also consistent with Detective Ross’ recollection of Ms. Freedhoffs account: As.H. and others “could be the people that robbed her,” and As.H. “looked like” one of the kids.
*461It is, of course, difficult (if not impossible) to place a meaningful numerical value on reasonable doubt. See generally Tribe, Trial by Mathematics: Precision and Ritual in the Legal Process, 84 HARV. L. REV 1329 (1971); Underwood, The Thumb on the Scales of Justice; Burden of Persuasion in Criminal Cases, 86 Yale L.J. 1299, 1309-11 (1977) (hereinafter Underwood ). Professor Wigmore cites a study in which judges in Chicago were asked to
translate into probability statements their sense of what it means to be convinced by a preponderance of the evidence, and to be convinced beyond a reasonable doubt. When responding to questionnaires, at least, the judges thought there was an important difference: almost a third of the responding judges put “beyond a reasonable doubt” at 100%, another third put it at 90% or 95%, and most of the rest put it at 80% or 85%. For the preponderance standard, by contrast, over half put it at 55%, and most of the rest put it between 60% and 75%.
J.H. WIGMORE, EVIDENCE IN TRIALS AT COMMON LAW § 2497, at 415 n.9 (1981) (quoting Underwood, 86 Yale L.J. at 1309-11; see also United States v. Fatico, 458 F.Supp. 388, 410 (E.D.N.Y.1978), aff'd, 603 F.2d 1053 (2d Cir.1979), cert. denied, 444 U.S. 1073, 100 S.Ct. 1018, 62 L.Ed.2d 755 (1980) (quoting the same study). Although the Chicago study alone is not dispositive of this appeal, it reveals that very few judges, if any, would have regarded an 80% probability as sufficient to prove guilt beyond a reasonable doubt, and that all of them would have considered a 70% probability as altogether inadequate. For the Chicago judges, Ms. Freedhoffs “certainty” appears to be well outside the ballpark for proof in a criminal case.8 In Fatico, 458 F.Supp. at 410, nine judges of the United States District Court for the Eastern District of New York, responding to a poll by Judge Weinstein, the co-author of a leading text on evidence,9 suggested percentages of 76%, 80%, 85%, 85%, 85%, 85%, 90%, 90% and 95%, as reflecting the standard for proof beyond a reasonable doubt. Thus, at most, two of the nine judges polled by Judge Weinstein would have found the level of assurance voiced by Ms. Freedhoff sufficient to support a finding of guilt.
But, argues the District, Ms. Freedhoff “was not asked for a level of accuracy or how sure she was, but, given the certainty of her identification, how high a level of certainty she had felt.” Therefore, the argument goes, “the trier of fact can be confident that the witness felt that her identification was very certain.” We do not find this contention at all persuasive. Taken to its logical conclusion, it would mean that if Ms. Freedhoff had expressed a level of certainty of one in ten — 10%— this would be sufficient to support a finding of guilt. The notion that Ms. Freed-hoff was assessing varying gradations of certainty, all of them very certain, is also *462at odds with what she told Detective Ross, namely, that As.H. “looks like” or “could have been” one of the robbers.10
Professor Lawrence Tribe has written: [I]t may well be ... that there is something intrinsically immoral about condemning a man as a criminal while telling oneself, “I believe that there is a chance of one in twenty that this defendant is innocent, but a 1/20 risk of sacrificing him erroneously is one I am willing to run in the interest of the public’s — and my own — safety.”
Tribe, supra, 84 HARV. L. REV. at 1872; see also Charles R. Nesson, Reasonable Doubt and Permissive Inferences: The Value of Complexity, 92 HARV. L. REV. 1187,1197 (1979).11 It may be that Professor Tribe’s proposition is more suited to the world of academe than to the less rarefied realities of the Superior Court’s criminal docket — realities in which “beyond all doubt” presents an idealist’s impossible dream, while “beyond a reasonable doubt” provides a workable standard. This case, however, is not like the hypothetical one that disturbed Professor Tribe. Here, the doubt of the sole identifying witness in a night-time robbery by strangers to her stood at two or three out of ten, or 20%-30%. We conclude, at least on this record, that this level of uncertainty constituted reasonable doubt as a matter of law. Accordingly, we reverse the adjudication of guilt and remand the case to the Superior Court with directions to enter a judgment of not guilty and to dismiss the petition.12
*463 So ordered. 13
.- As.H. was born on April 29, 1982.
. Ms. Freedhoff testified that there were three assailants. According to Detective Michael Ross of the Metropolitan Police Department (MPD), however, Ms. Freedhoff initially re*458ported that there were "anywhere from three to five” suspects involved in the robbery.
. In his affidavit in support of a request for a "custody order” (i.e., the juvenile counterpart of an arrest warrant), Ross quoted the complainant as saying that the photograph of As.H. "looked like one of the kids.”
. On appeal, As.H. contends that his motion to suppress was erroneously denied. In light of our disposition of the case on other grounds, we need not address this contention.
. Or, in this case, in a juvenile delinquency proceeding.
. Or, in this case, a trial judge.
. The court emphasized in Crawley that, as appellate judges, we have the responsibility in eyewitness identification cases "to draw upon our own experience, value judgments, and common sense in determining whether the [finding] reached was in keeping with the facts.” Id. at 312. Although this observation might be viewed today as an unduly activist formulation of an appellate court's function, it illustrates the concern of conscientious judges regarding the possibility that a mistaken identification may send an innocent person to prison.
. Commenting on the same Chicago study in one of its submissions, the District reveals only that "about one-third of the judges put it at 80%-85%.” Unfortunately, by failing to mention that one third of the judges put "beyond a reasonable doubt” at 100% and that another third put it at 90%-95%, the District presents us with a misleading picture of the results of the study. Remarkably, the District then argues that we should affirm because judges who try to quantify reasonable doubt place it “not that far distant from Ms. Freed-hoff’s estimate.” In fact, the contrast between the judges’ estimates and Ms. Freed-hoff’s articulation is quite remarkable, and a contention that fails to take this contrast into account is necessarily fallacious.
. JACK B. WEINSTEIN AND MARGARET A. BERGER, WEINSTEIN'S FEDERAL EVIDENCE (2d ed.2004).
. The District also argues that, in open court, Ms. Freedhoff “unhesitatingly and positively” identified the respondent. As we have explained in Part I of this opinion, however, the full context of Ms. Freedhoff's courtroom testimony reveals that, five months after the robbery, she was no more certain of her identification than she had been when she viewed the photo array.
Moreover, after Ms. Freedhoff had selected photographs at the array, Detective Ross revealed that he "had a previous history with the persons she had picked,” and that they were his "possible suspects in the case.” "[W]here ... the police consider an individual to be a possible perpetrator and a witness makes an initially ambiguous identification, there may develop a process of mutual bolstering which converts initial tentativeness into ultimate certainty.” In re Dwayne W., 109 Daily Wash. L. Rptr.1901, 1906 (Super. Ct.D.C, 1981). "The victim relies on the expertise of the officer and the officer upon the victim’s identification.” Id. (quoting [JUDGE] NATHAN SOBEL, EYEWITNESS IDENTIFICATION, LEGAL AND PRACTICAL PROBLEMS 12 (1972 & Supp.1981)).
. See also Maimonides, Safer HaMitzvot, Negative Commandment 290, quoted in Fático, 458 F.Supp. at 410, for one religious assessment of the issue:
The 290th commandment is the prohibition to carry out punishment on a high probability, even close to certainty .... If we do not punish on very strong probabilities, nothing can happen other than that a sinner be freed; but if punishment be done on probability and opinion it is possible that one day we might kill an innocent man — and it is better and more desirable to free a thousand sinners, than ever to kill one innocent.
.Our dissenting colleague argues that reasonable doubt is not susceptible of ready quantification, and we agree. But where, as in this case, the sole identifying witness described her own level of "certainty” as only seven or eight on a scale of ten, then, notwithstanding the difficulty of quantification in the abstract, this level of unsureness necessarily raises a reasonable doubt and negates the requisite finding of "near certitude” that As.H. was one of the robbers.
Nothing in this -opinion holds or even remotely suggests that a cross-racial identification is insufficient as a matter of law or that the trier of fact is required to discount such an identification. The reasonable doubt in this case arises from the witness’ very limited certainty (seven or eight on a scale of ten) regarding her uncorroborated identification. The difficulties of eyewitness identification of strangers in general, as well as of cross-racial *463identification, provide the context in which the witness’ uncertainty arose.
. Although the parties did not cite them, we have found several decisions that skirt the issue in this case, but none is especially helpful. See Rustin v. State, 46 Md.App. 28, 415 A.2d 631, 635 & n. 2 (1980) (testimony that the appellant "resembled” one of the robbers apparently inadmissible; 95% certainty, however, was admissible; "|l]ack of certainty is a factor weighing against the admissibility of the identification”); People v. Giocastro, 210 A.D.2d 254, 619 N.Y.S.2d 354, 355 (1994) (per curiam) (where, inter alia, witness was "only 80%” certain of his pretrial identification almost two months after offense, and where he "hesitated before identifying the defendant in court,” the proof, though "legally sufficient,” was "against the weight of the evidence,” and the court "ha[d] a reasonable doubt” and held that "the risk of misidentifi-cation is too great to allow the conviction to stand”); Orman v. State, 37 A.D.2d 674, 322 N.Y.S.2d 914, 916 (1971) (per curiam) (dictum) (in civil case governed by "preponderance of the evidence” standard, testimony of witness who forthrightly stated that he was "only 75%-85% sure of his identification" apparently deemed sufficient; judgment for defendant on other grounds).