Following a joint jury trial on charges arising out of two separate incidents, appellants Gary Byrd and Terrence Lewis each were convicted of two counts of assault with a dangerous weapon, D.C.Code § 22-502 (1981), one count of assault with intent to kill while armed, id. § 22-501, and two counts of carrying a pistol without a license, id. § 22-3204. In these consolidated appeals, appellants primarily contend that the offenses arising out of the respective incidents were misjoined under Super. Ct.Crim.R. 8(b).2 While we agree that the offenses were improperly joined, we conclude that the error was harmless. Accordingly, we affirm all convictions.

I

The government’s evidence showed that on January 18, 1984, at approximately 9:00 p.m., James Van proceeded homeward along Lincoln Road, N.E. As he reached the corner of Lincoln Road and Randolph Place, he was approached by two men who asked him for a cigarette. Van replied that he did not have one, and the men then drew guns and announced a robbery. After threatening to shoot him, the assailants threw him down into a lot. They searched his pockets, took about twelve or thirteen dollars, and then shot him three times in his left leg and once in his right leg. At trial, Van testified that one of the guns used by the robbers was steel blue in color, while the other was nickel-plated with a white, possibly pearl, handle.3

Timothy Smith, standing nearby, heard the gunfire and ran up S Street where he saw Van fall to the ground. Smith then ran to the comer of Lincoln Road and S Street and saw two men running toward what he described as a sky-blue Lincoln Versailles with spoke-rim wheels. Although Smith viewed both men only from the back, he recognized one of them as Terrence Lewis.4 Later that night, Smith saw the same car at North Capitol and K Streets where it had been stopped by the police, and thought he saw Lewis, although again only viewing him from the back.5

*98That same night, at approximately 10:20 p.m., Odell Hilman, Sylvester Moore, and a young woman named Yetta, were standing at the corner of 4th Street and Rhode Island Avenue, N.E. As Odell exited a convenience store, he bumped into Lewis. Odell then rejoined Moore, and Lewis got into a blue Lincoln Versailles with spoke-rim wheels. The car drove off but returned fifteen minutes later. Before the car returned, however, Yetta went home, and Odell and Moore were joined by Odell’s brother Wayne. When the car finally returned, Byrd, who was sitting in the rear passenger seat, yelled out a question as to the whereabouts of Yetta. Odell responded that Yetta had gone home. Byrd then motioned Odell toward the car. When Odell approached, he recognized Lewis, who was sitting in the front passenger seat, as the same individual he had bumped into earlier. After a brief verbal exchange, Byrd drew a gun and pointed it at Odell. A struggle ensued, and both Byrd and Lewis pulled Odell by his coat until he was pinned against the car. The altercation ended when Byrd pointed the gun at Odell and fired. As the car drove off, Byrd fired two more shots, hitting Wayne in the right arm.6

At about midnight, Detective Thomas and his partner Officer Cianciotti spotted a car at the 100 block of Rhode Island Avenue, N.E., which closely matched the description of the vehicle being broadcast as the one used in the Hilman shootings.7 As the officers followed the car, it increased speed. In response the officers turned on their emergency siren and red light, but the car did not stop. Finally, the car came to a halt in the middle of the road. As Cianciotti approached, it sped off, but was finally stopped at K and North Capitol Streets with the assistance of several other officers.

When the officers asked both men to identify themselves, Byrd, who was sitting in the driver’s seat, produced a license bearing the name of another individual. Lewis, who was sitting in the front passenger seat, also identified himself as someone else. A search of the trunk of the vehicle uncovered two weapons: a silver .32 caliber Colt semi-automatic pistol upon which a palm print of Byrd was found, and a blue, somewhat rusted, .38 caliber Rolm revolver. Ballistics evidence indicated that a .32 caliber shell casing found at the scene of the Van shooting had been fired from the .32 caliber Colt revolver, and that a .38 caliber slug removed from Van had been fired either from the .38 Rolm revolver or an essentially identical type of gun. Odell Hilman testified that the .32 caliber pistol resembled the weapon he was shot with.

At trial, both appellants offered alibi defenses placing them away from the scenes of the incidents at the relevant times.

II

Appellants contend that the charges relating to the first incident were improperly *99joined with the charges relating to the second, and therefore the trial court erred in denying their motions to sever offenses. Super.Ct.Crim.R. 8(b) permits the joinder of offenses in a multiple defendant case, but only if the offenses “are based on the same act or transaction or series of acts or transactions.” Ray v. United States, 472 A.2d 854, 857 (D.C. 1984). Multiple offenses are said to constitute the same act or transaction or series of acts or transactions: “(1) where the offenses are committed to achieve a ‘specific common end,’ (2) ‘where one offense logically leads to another,’ or (3) ‘where the offenses are part of a common scheme or plan,’ and are so closely connected in time or place ‘that there is necessarily a substantial overlap in proof of the various crimes and it would be difficult to separate proof of one from the other.’ ” Settles v. United States, 522 A.2d 348, 352 (D.C. 1987) (quoting Davis v. United States, 367 A.2d 1254, 1262 (D.C. 1976)).

On the facts before us we conclude that the two sets of offenses were not based on the same act or transaction or series of acts or transactions and were therefore improperly joined under Rule 8(b). Clearly, there was no specific common end toward which the two sets of offenses were directed. Each incident comprised an isolated event, and neither depended for its furtherance or success upon the other. Id. at 353 (citation omitted). In addition, the Hilman incident was not a continuation of the Van incident and “in no way did it logically or necessarily result from” that incident. Id, Thus, there was “no logical development of or relationship between the offenses.” Davis, supra, 367 A.2d at 1263. And, finally, there was not substantial overlap of proof between the two sets of offenses. Although both crimes were very closely connected in time and place, and there existed some overlap of evidence, it was not difficult to separate proof of one crime from that of the other. Settles, supra, 522 A.2d at 353. At trial, the government presented its proof separately and distinctly, and there was no overlap of testimony by a government witness from one incident to the other. Id. at 353-54 (citation omitted). Thus, there did not necessarily exist a substantial overlap of proof between the respective offenses, and, in turn, there was “no evidentiary need for joinder.” Id. at 353.

Ill

We must next determine, however, whether the misjoinder was harmless error. Settles, supra, 522 A.2d at 354 (citing United States v. Lane, 474 U.S. 438, 106 S.Ct. 725, 88 L.Ed.2d 814 (1986)). Misjoinder may be deemed harmless when it does not result in actual prejudice, which is to say “only if it has no ‘substantial and injurious effect or influence in determining the jury’s verdict.’ ” Id. at 354 (quoting Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946)). We have construed this language to mean that “a misjoinder may be deemed harmless only if ‘all or substantially all of the evidence of one offense would be admissible in a separate trial of the other.’ ” Id. at 354 (quoting Ray, supra, 472 A.2d at 859 (citations omitted)).8 In determining whether offenses are mutually admissible, we apply the principles articulated in Drew v. United States, 118 U.S.App.D.C. 11, 331 F.2d 85 (1964). Drew permits the introduction of other crimes evidence when relevant to motive, intent, absence of mistake or accident, common scheme or plan, or identity, and when the probative force of the evidence outweighs its inherently prejudicial effect. Id. at 16, 331 F.2d at 90. When offenses are mutually admissible, “there can be no danger of prejudice in trying the cases together rather than separately ‘because in either event the jury will *100hear all about both crimes.’ ” Settles, supra, 522 A.2d at 354 (citation omitted). Joinder is then permitted “despite [the presumptive] possibility of prejudice due to the cumulation of evidence against the defendants, because it will ‘promote economy and efficiency and ... avoid a multiplicity of trials.’ ” Ray, supra, 472 A.2d at 859 (quoting Davis, supra, 367 A.2d at 1263).

Upon review of the record, we conclude that evidence of either crime would have been admissible in a separate trial of the other in order to corroborate the circumstantial evidence pointing to the participation of appellants as the perpetrators of both crimes. Thus, in effect, the core issue is one of identification.

The first step in the analysis is to focus on the logical relevance issue in depth in order to isolate the inferences by which the identity of the perpetrator may be proved. Our test “is to determine if there are enough points of similarity in the combination of circumstances surrounding the two crimes to create a reasonable probability that the same person[s] committed each.” Easton v. United States, 533 A.2d 904, 908 (D.C. 1987). The test, however, is one of reasonable probability, not of scientific certainty.9 See Easton, supra, 533 A.2d at 908-09.

Turning to the instant case, we conclude that the combination of circumstances surrounding the commission of both crimes creates a reasonable probability that the same persons committed each. Of particular significance is the fact that the crimes occurred within ninety minutes of each other at locations approximately six blocks apart. See Gates v. United States, 481 A.2d 120, 123 (D.C. 1984) (crime occurring nineteen days apart and at locations within a few hundred yards); Brooks v. United States, 448 A.2d 253, 257 (D.C. 1982) (crimes occurring nine days apart); Cox v. United States, 498 A.2d 231, 238 (D.C. 1985) (crimes occurring two days apart and in same general wooded area); Bowyer v. United States, 422 A.2d 973, 977 (D.C. 1980) (three of four incidents occurring within four to six weeks of each other); Samuels v. United States, 385 A.2d 16, 19 (D.C. 1978) (crimes occurring seven hours apart); compare Tinsley v. United States, 368 A.2d 531, 536 (D.C. 1976) (evidence not admissible noting that crimes occurred six months apart and at distant locations). Each crime also involved the use of a distinct type automobile, the blue Lincoln Versailles with spoke-rim wheels, which proved to be the same type vehicle in which appellants were driving when stopped by the police less than two hours after commission of the second crime. See Warren v. United States, 436 A.2d 821, 832 (D.C. 1981) (small green sportscar with black interior and bucket seats); Bowyer v. United States, 422 A.2d 973, 977 (D.C. 1980) (dark blue Thunderbird).10 Finally, a silver .32 caliber pistol, upon which a palm print of Byrd was found, was conclusively tied to the Van incident through ballistics evidence and identified by Odell Hilman as the weapon he was shot with in the other incident. See Cox, supra, 498 A.2d at 238 (same type gun).

Of course, the question of legal relevance must also be considered. “When evidence is relevant and important to [the issue of identity], its probative value is deemed to overshadow an otherwise impermissible prejudicial effect and, thus, the evidence of each crime is admissible at a separate trial of the other offense.” Tinsley v. United States, 368 A.2d 531, 534 *101(D.C. 1976). In the instant case, the identities of the perpetrators of both incidents were clearly in dispute, and each appellant offered an alibi defense based on a theory of misidentification. Evidence of the second crime would have been admissible in a separate trial of the first in order to corroborate what were plausible but far from overwhelming identifications of both appellants.11 Evidence of the first crime would have been admissible in a separate trial of the second for the same reason.12 Therefore, introduction of evidence of either crime in a separate trial of the other would have “promise[d] ‘a real contribution in the process of proof,’ ” Easton v. United States, supra, 533 A.2d at 906, and thus would have proved more probative than prejudicial.

In sum, we are satisfied that evidence of either crime would have been admissible in a separate trial of the other in order to corroborate the circumstantial evidence pointing to the participation of appellants as the perpetrators of both crimes. Our conclusion follows then that the misjoinder in the instant case was harmless error.

Affirmed.

dissenting:

In this case, I cannot agree with the majority that the misjoinder of separate offenses for trial, in violation of Super.Ct. Crim.R. 8(b), (and the denial of a motion to sever by the trial court) nevertheless constituted harmless error. Although misjoin-der is not reversible error per se, a presumptive possibility of prejudice to the defendant results from improper joinder under Rule 8(b). Morris v. United States, 548 A.2d 1383 (D.C. 1988).

The majority not only ignores this “presumptive possibility” but makes short shrift of a complex task. Conceding that the evidence of guilt in this case is not overwhelming, and finding it unnecessary to consider the effectiveness of the trial court’s instructions to the jury, the majority focuses solely on a mutual admissibility theory {see Settles v. United States, 522 A.2d 348, 354 (D.C. 1987) (citing Ray v. United States, 472 A.2d 854, 859 (D.C. 1984) (“a misjoinder may be deemed harmless only ‘if all or substantially all of the evidence of one offense would be admissible in a separate trial of the other’ ”)). The evidence as to one crime, says the majority, would be admissible in a trial for the other crime to prove identity. Thus, the majority slides into a thicket of misconceptions with respect to other crimes evidence. See Drew v. United States, 118 U.S.App.D.C. 11, 331 F.2d 85 (1964); see also Thompson v. United States, 546 A.2d 414 (D.C. 1988); Bartley v. United States, 530 A.2d 692, 701 (D.C. 1987) (Mack, J., dissenting).

*102In the typical Drew analysis, the instant case does not present a situation in which the evidence of one crime would have been admissible in a separate trial of the other to prove identity. It would not have been enough to have simply alleged that identity was an issue. There must have been clear and convincing evidence that the defendant was the perpetrator or co-perpetrator of the uncharged crime. See Light v. United States, 360 A.2d 479, 480 (D.C. 1976); United States v. Bussey, 139 U.S.App.D.C. 268, 273, 432 F.2d 1330, 1335 (1970). There must have been an articulated independent theory of logical relevance — how or by what means the uncharged crime would have proved identity. See Ali v. United States, 520 A.2d 306, 310 n. 4 (D.C. 1987); Drew, supra, 118 U.S.App.D.C. at 16, 331 F.2d at 90. The proof of modus operandi must have been of such nature as to make it highly likely that the same person committed both crimes. A modus operandi which was not unique would not have been sufficient. Easton v. United States, 533 A.2d 904, 907 (D.C. 1987); Warren v. United States, 436 A.2d 821, 832 (D.C. 1981). Finally there would have had to have been a careful and informed exercise of discretion by the trial court finding that the probative effect of the evidence would have outweighed the prejudice to the accused. See Graves v. United States, 515 A.2d 1136, 1139 (D.C. 1986).

Applying these relevant factors here, I do not see how this court can conclude that the evidence of one crime would have been mutually admissible at the trial of the other and that therefore the misjoinder was harmless error. The majority may be right in concluding that there was a need to corroborate the circumstantial evidence as to identity, which it concedes is not overwhelming, but that is simply another reason for finding that the misjoinder could not have been harmless. A theory of logical relevance supporting admissibility is lacking; the majority chooses not to speak in terms of modus operandi but alludes to the “points of similarity surrounding the two crimes to create a reasonable probability that the same person[s] committed each.” We are not told why the points of similarity or combination of circumstances are unique. There is nothing unique about violent crimes occurring within ninety minutes of each other in an urban setting. There is nothing unique about a .32 caliber pistol or a blue automobile with spoke-rim wheels (which the majority concedes was described with varying degrees of particularity). If there are similarities, they are superficial and unremarkable.

The truth of the matter is that the majority, in holding that there was a misjoinder of offenses under Rule 8(b), has placed itself on the horns of a dilemma. In its effort to find this misjoinder harmless, it has embraced a mutually admissible evi-dentiary theory which highlights the fact that the misjoinder could not have been harmless, i.e. that the challenged evidence had a “substantial and injurious effect or influence in determining the jury’s verdict.” Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946); see United States v. Lane, 474 U.S. 438, 449, 106 S.Ct. 725, 732, 88 L.Ed.2d 814 (1986).

In its most basic sense, the determination of whether an error in the admissibility of evidence is nevertheless harmless is not an easy one. It is made more difficult when the challenged evidence is other crimes evidence which Drew mandates for exclusion or permits as an exception. When the asserted exception is one of identity, and when there is nothing of record to tie the defendant to the other crime sought to be introduced, the risk to the defendant in the reception of such evidence is monumental. The task of sorting out the degrees of similarity between the commission of a charged crime and the other crime may be a tedious one. See Easton v. United States, supra, 533 A.2d at 908. If the modus operandi of the two crimes is not so unique as to make it likely that the same person committed both crimes, the prejudice to the accused is apparent, particularly if the challenged evidence constitutes a substantial part of (or is critical to, as here) the government’s case. Id. at 909. Given all these difficulties in weighing probative value and prejudicial effect, for the pur*103pose of assessing harmless error, still another dimension is present when the fact of improper joinder is added to the Drew calculus. This court has noted that there has been some disagreement among courts on whether a misjoinder under Rule 8(b) could ever amount to harmless error. Settles v. United States, supra, 522 A.2d at 348. The circumstances of the instant case illustrate the magnitude of this problem.

Thus, there is a certain incongruity in pronouncing, as has the majority here, that misjoinder was error because there was no evidentiary need for joinder, and at the same time proclaiming that, given the need for identification evidence, mutual admissibility would have provided a real contribution to the process of proof. Perhaps that is why this court, while recognizing the applicability of the harmless error analysis to misjoinder, has been reluctant to find harmless error as to Rule 8(b) violations.1 See Morris v. United States, supra; Easton v. United States, supra; Settles v. United States, supra; Ray v. United States, supra; Tinsley v. United States, 368 A.2d 531 (D.C. 1976); Davis v. United States, 367 A.2d 1254 (D.C. 1976). Cf. Wright v. United States, 510 A.2d 223 (D.C. 1986), where the evidence of the defendants’ involvement in the distribution of heroin was “compelling” and where a protective instruction was given.