delivered the opinion of the court:
The prosecution charged defendant, Joecephus Mitts, with three separate sexual assaults that occurred in the Englewood area in an eight-day period. The jury found him guilty of two of the assaults, but not guilty of the third. Defendant appeals from his convictions and sentences.
Just before 7 a.m. on December 14, 1993, L.B. walked down Hoyne Street towards a bus stop on 55th Street. A man across the street, walking the opposite direction, began crossing the street just as she started to cross. As she walked past, the man grabbed her arm and said, “[T]his is a stickup.” He pulled out a gun and pointed it at her head. He told her to walk with him, and if she ran he would shoot her. He told her to take out her money. She did not give him any money.
The man took L.B. to a garage in a nearby alley and told her to take off her clothes. He threw her to the floor and lay on top of her, with the gun pointing to her head. He had vaginal intercourse with her. He struck her face with his fist and told her she cried too much. Then he left.
Once the man was out of sight, L.B. ran home and her family called the police. She described the assailant as a black male, about 5 feet 5 inches and 150 pounds, wearing a black, hooded sweatshirt, blue jeans, gym shoes, and a knit cap. Police took L.B. to a nearby hospital, where they obtained a swab of semen found in L.B.’s vagina.
*4Two days later, around 1:30 a.m., T.S. passed a man in an alley as she walked on Wolcott Avenue, a few blocks from the location of the attack on L.B. The man walked up behind her, put a gun to her back and told her it was a stickup, and if she said anything he would “pop” her. He pulled her by the arm through a gangway and into a garage. He kept repeating that if she said anything he would “pop” her.
Once in the garage the man told T.S. to give him her money. When she said she had none, he said he would get something from her. He told her to get on her knees, and he put his penis in her mouth. He then told her to take off her clothes. He took a neck chain from her and told her to bend over the car. He had vaginal intercourse with her. He told her to wait five minutes or he would “pop” her. She waited, then she ran home and her family called the police.
T.S. described the assailant as a black man, around 20 years old, 5 feet 6 inches and 180 pounds, wearing a black Starter jacket, dark pants, and a skull cap. Police took T.S. to a nearby hospital, where they obtained a swab of semen found in her vagina.
Shortly after midnight on December 22, 1993, L.W got off a bus on 55th Street and walked down Damen Avenue, less than two blocks from the location of the attack on L.B. A man walked up behind L.W and said, “[T]his is a stickup, bitch.” He grabbed her neck and pointed a gun at her back. He pushed her to a gate and told her to climb over it. He climbed over and took her into a garage. He pushed her to her knees and said “stay there.” He put his penis in her mouth. Then he told her to turn around. He pulled down her pants and made her lie on her jacket. He put his penis first in her vagina and then in her anus.
When the man got up, he asked L.W whom she knew. She gave the name of a relative, and the man said he “fucked with the wrong female.” He recognized the relative as a member of the Blackstones gang. He told L.W. to stay in the garage until he was gone. Once he left she ran home, where her family called police.
L.W. described the man who attacked her as 17 to 24 years old, about 5 feet 8 inches and 170 pounds, wearing a black cap pointed left, a black, hooded coat, black pants and black shoes. Police took L.W. to a hospital where they obtained a vaginal swab including semen.
After midnight on December 23, 1993, Officer Terrence Johnson patrolled the area where the crimes occurred, looking for persons similar to the descriptions L.B., T.S. and L.W. gave of the offenders. He saw defendant and two other black men talking near the corner of 55th and Seeley, which is between Damen and Hoyne. As he approached, the other two men began walking east on 55th. Johnson asked defendant his name, where he lived, and what he was doing on *5the corner. Defendant gave his name and address and told the officer he was going to the store. Johnson then went to ask the same three questions of the other two men.
Defendant was 23 years old, 5 feet 6 inches and 190 pounds. Because Johnson thought defendant fit the descriptions of the offenders fairly well, he requested a photograph of defendant from police department files. The process of obtaining the photograph took several days.
On December 28, 1993, Sergeant Kevin Duffin watched defendant walking near 55th and Damen. Duffin spoke to defendant briefly, searched him but found no weapon, then took him to the police station before driving defendant back to Englewood. At some point Duffin asked defendant some questions.
When Johnson obtained defendant’s photograph on December 30, 1993, he assembled a photo array and brought it to L.B. L.B. picked defendant’s photograph as a picture of the man who attacked her. Police arrested defendant and put him in a lineup. Both L.B. and L.W identified defendant as the man who assaulted them. In a later lineup T.S. also identified defendant as the man who attacked her.
In May 1994 the police lab compared the DNA found on the swab taken from L.B.’s vagina with the DNA from L.B.’s and defendant’s blood. The lab found a match with defendant at six separate locations where DNA varies sharply from person to person.
The lab also compared the DNA found on the swab of L.W. with L.W.’s and defendant’s blood. The DNA on the swab matched neither L.W. nor defendant at most locations, but at some locations the analyst, Pamela Fish, found faint bands that might possibly match defendant’s DNA, along with much clearer bands that matched neither defendant nor L.W. In September 1994 Fish told the assistant State’s Attorney to look for a friendly source responsible for the clear bands matching neither L.W. nor defendant.
The assistant State’s Attorney never responded to Fish and never requested testing of the swab from T.S. The prosecution supplied the DNA tests to defense counsel but never informed the defense of the identification of a friendly source for the DNA found on the swab from L.W
The prosecution moved to have evidence pertaining to all three assaults admitted into the case concerning the assault on L.B. The court found sufficient indication of a single perpetrator using a distinctive modus operandi, and therefore the court granted the prosecutor’s motion. The court subsequently permitted the prosecution to try defendant on charges of committing the three assaults in a single trial.
The defense moved for a Frye hearing concerning the admissibility *6of DNA evidence. In support of the motion defendant presented statements and articles from several experts concerning testing procedures and statistical analyses of results. The experts criticized the lack of random sampling in the creation of DNA databases used for computing the probability of random matches. They also criticized police labs for their failure to perform blind proficiency tests to check on the accuracy of procedures. None of the police labs reported error rates. In forensic tests, the analysts always know in advance which samples police want the lab to match, contrary to good scientific technique.
In one of the articles submitted in support of the motion, Harvard University professor R.C. Lewontin explained the problem of laboratory error when the lab receives a small sample from a crime scene and some source to compare, like blood, from a suspect.
“While there is more than enough DNA recoverable from the suspect’s large blood sample to carry out the needed procedures, the very small, and often degraded, sample from the crime scene does not contain sufficient DNA for the comparison. To obtain sufficient material, the DNA from the crime scene is ‘amplified,’ that is, copied thousands or millions of times in a procedure known as polymerase chain reaction (PGR). *** The problem with the PGR technique is that because of its chain nature, contaminant molecules in the original sample may also be amplified and, since the original crime scene sample contained so few molecules, contaminants may overwhelm the original in the amplification. *** Now consider the actual practice in a forensic DNA laboratory. A technician is handling two samples. One is the very large DNA sample from the suspect’s blood, the other is the minuscule DNA sample from the crime scene, which is then amplified by PGR. The situation is ideal for PGR contamination, with the result that the suspect’s DNA will not really be compared with that from the crime scene, but with his or her own DNA that has just been replicated in the PGR reaction. The result will be a perfect match.
All of us who use the PGR technique regularly are acutely conscious of the contamination problem, and the best laboratories have suffered occasionally from it. The perspiration and ‘oils’ on fingertips have provided enough DNA contamination in PGR experiments to give completely artefactual results.” R. Lewontin, Comment: The Use of DNA Profiles in Forensic Contexts, 9 Stat. Sci. 259 (1994).
One study of laboratories in California found 2 false positives reported from 110 samples tested. Based on the limited data available, an expert said “a reasonable estimate of the false positive error rate is 1-4 percent.” J. Koehler, DNA Matches and Statistics: Important Questions, Surprising Answers, 76 Judicature 222, 229 (1993).
*7The prosecution presented no evidence in response. The court denied the motion for a Frye hearing and decided to admit the prosecution’s DNA evidence, along with testimony from a defense expert.
Defendant moved to suppress identification testimony, arguing first that Officer Johnson violated his right to be free from unreasonable seizure when Johnson stopped defendant on December 23, 1993. The court rejected the argument. Defendant also argued that the lineups were unduly suggestive. The court looked at the photographs of the lineups and agreed with defendant that the lineups were not perfect. Defendant was the darkest person in the first lineup, the shortest in the second lineup, and one of the heaviest in each lineup. But the court held that the lineups “pass constitutional muster.” Accordingly, the court denied the motion to exclude identification evidence.
Defendant moved to suppress statements he allegedly made to Sergeant Duffin. Duffin testified that on December 28, 1993, he watched defendant walk back and forth near 55th and Damen for three hours. Duffin then approached and asked defendant who he was and why he was out on the street on such a cold night. Defendant gave his name and said he was just going to a nearby mini-mart. Duffin asked to see some identification, but defendant did not have any. Defendant said he thought there might be a warrant outstanding for him. He explained that Evergreen Park police had arrested him for possession of a controlled substance, and he might have missed the scheduled court date. Duffin offered to drive defendant to the police station to determine whether the court had issued a warrant for his arrest. Defendant agreed. Although he searched defendant before letting him enter the police car, Duffin found no weapon.
Duffin testified that in the car he talked with his partner about handguns. Defendant asked what kind of guns they carried. After Duffin answered, he asked defendant whether he ever owned a gun. Defendant answered that he had owned a revolver, but he sold it the prior week. At the station, police determined that defendant had not missed his court date on the possession charge. Duffin drove defendant back to the location from which he picked him up.
Defendant testified that police arrested and handcuffed him just after he got off a city bus. An officer took his identification and put him in the back of a police car. The police did not ask him anything in the ride to the station, and defendant asked them only why they arrested him. They questioned him at the station, without Miranda warnings, for half an hour before driving him home. Defendant still said nothing about guns.
*8Based on his appraisal of the credibility of the witnesses, the judge denied the motion to suppress the statement. The trial began in October 1997.
After jury selection the prosecutor informed defense counsel that L.W would testify she had consensual sex with her boyfriend a few days before the attack. Defendant moved for a mistrial so that the defense could find the boyfriend and test his DNA to determine whether it matched the DNA found on the swab of L.W. The prosecutor made no response. No evidence suggested that LW was ever unavailable for questioning, and the prosecutor knew of the need to question her about a boyfriend at least three years before trial. The court denied the motion, thereby requiring defendant to go to trial without the ability to compare the boyfriend’s DNA to the DNA on the swab.
L.B., T.S. and L.W. separately identified defendant in court as the man who attacked each of them. L.W. testified that she had intercourse with her boyfriend two or three days before the attack.
The analyst who tested the semen found in the attack on L.B. swore defendant’s DNA matched the DNA on the swab at six locations, and based on the lab’s DNA database she estimated the probability of a random match at less than one in one billion.
Pamela Fish, who tested the semen found after the attack on L.W., testified that some semen may remain traceable in the vagina for 72 hours after intercourse. She admitted that the sperm most clearly present in L.W did not match defendant at all, but she found faint traces indicative of sperm from a second source. At one location on the DNA she found the faint band matched defendant’s DNA, and at other locations the traces were far too light to positively rule out defendant as a potential source. The prosecution never asked her to compare the swab from L.W. with her boyfriend’s DNA. The analyst admitted that the light banding pattern showed considerable signs of degradation. She also admitted that the swab contained far more DNA which matched neither LW nor defendant.
On cross-examination, the defense asked the prosecution’s expert about the composition of the database and whether the lab had procedures to assure randomness to give an accurate picture of the DNA available in the population as a whole:
“Q. Those data bases, the vast majority, 90-some percent, of the samples you got came from the Chicago Morgue, right?
A. Correct.
Q. Basically they were murder cases, correct?
A. I can’t say what all of the samples came from, but most of them did come from the Cook County Morgue, the Medical Examiner’s Office.
*9THE COURT: The morgue handles homicide cases and it handles natural death cases and it handles accident cases.
* * *
Q. In fact, if you go out and look at the actual files associated with those case numbers [of the samples that come to the lab for inclusion in the database], the vast majority of them are murder cases, right?
[Prosecutor]: Judge, I object.
THE COURT: Sustained. ***
Q. *** [I]n order for these data bases to be valid data bases they have to be randomly selected, right?
A. Correct.
Q. Essentially, the murderers in Chicago were doing the selection, weren’t they?
[Prosecutor]: Objection.
THE COURT: Sustained. First, we are talking about victims, not murderers, from the morgue. And secondly, I have already indicated the jury can take note that the morgue treats accidental deaths and it treats natural deaths where the cause of death is not apparent. It treats all kinds of various death situations, and it doesn’t make any difference.”
At that point the defense abandoned the effort to show that department procedures did not meet scientifically acceptable standards for construction of a random database that accurately reflects the DNA available in the population.
Defendant’s expert testified that the faint bands shown on the test of the swab from L.W most likely arose from bacterial contamination or degradation of the DNA. At most locations, nothing at all appeared where banding should appear if DNA from two separate semen sources accounted for any of the banding. The evidence excluded defendant as a possible source of the semen. The expert described the evidence as “one of the cleaner exclusions [he had] ever seen.”
In response to the theory that the primary bands came from L.W.’s boyfriend, and the faint bands came from defendant, the expert noted studies showing that vaginal swabs taken 48 hours after intercourse do not provide any reliable material for DNA testing. If the banding pattern reflected two separate sources, then the clear banding pattern came from the more recent source, the man who attacked L.W., and the faint bands must reflect the DNA pattern of the semen left two to three days before the attack, when L.W. had intercourse with her boyfriend.
The jury found defendant not guilty of the attack on L.W., but guilty of aggravated criminal sexual assaults on L.B. and T.S., as well *10as armed robbery of T.S. and attempted armed robbery of L.B. The judge sentenced defendant to consecutive sentences totaling 210 years. The prosecution concedes that the sentence violates statutory limits of the court’s sentencing authority. See 730 ILCS 5/5 — 8—4(c)(2) (West 1996).
I
On appeal defendant argues that the State’s discovery violation requires reversal. The prosecution did not inform defendant until the day of trial, after jury selection, that L.W. had intercourse with her boyfriend two or three days before the attack.
The prosecution answers first that it did not learn of the evidence significantly before the defense learned it. However, nothing in the record supports the assertion. The laboratory analyst working with the DNA samples asked the prosecutor, three years before trial, to find out whether L.W had consensual sex shortly before the attack. No evidence showed that LW was unavailable to prosecutors for any significant time in that three-year period. Thus, the prosecution should have discovered the suppressed evidence approximately three years before trial, and no evidence shows a failure to discover it at that time.
Next, the prosecution contends that defendant waived objection to the discovery violation by failing to request a continuance rather than a mistrial. We disagree. A defendant’s failure to request a continuance does not waive objection to a discovery violation if a continuance cannot correct the prejudice. See People v. Weaver, 92 Ill. 2d 545, 559-60, 442 N.E.2d 255 (1982); People v. Matthews, 299 Ill. App. 3d 914, 921, 702 N.E.2d 291 (1998). A simple interview with L.W. or her boyfriend would not undo the damage to the defense case due to the discovery violation here. The defense needed to test the boyfriend to verify its expert’s theory that the boyfriend matched, at best, the faint banding the prosecution’s analyst found. The expert believed that the most evident DNA came from the man who attacked LW, and that man probably also attacked T.S. and L.B. The necessary testing would take far longer than the time for which the court could reasonably keep the jury waiting. Therefore, only a mistrial would give the defense the opportunity to perform the needed tests. Defendant properly preserved objection to the discovery violation when he sought appropriate relief at trial.
The prosecution also argues that defendant waived the issue by failing to raise it with sufficient specificity in the posttrial motion. In the motion defendant specifically objected to the court’s ruling on the motion for mistrial. The objection appears to be sufficiently specific, in *11the context of this case, to remind the trial court of the error the defense sought to rely upon, which is the same error he relies upon on appeal. See People v. Groves, 294 Ill. App. 3d 570, 574, 691 N.E.2d 86 (1998). Moreover, the waiver rule is a limitation on the parties and not on the jurisdiction of the court. People v. Shaw, 186 Ill. 2d 301, 327, 713 N.E.2d 1161 (1998). We choose to address the issue in part because defendant could raise waiver of the issue as grounds for finding ineffective assistance of counsel in a postconviction petition. See People v. Mitchell, 152 Ill. 2d 274, 285, 604 N.E.2d 877 (1992).
“Illinois courts have relied on the following factors to determine whether a defendant is entitled to a new trial as a result of a discovery violation: the closeness of the evidence, the strength of the undisclosed evidence, the likelihood that prior notice would have helped the defense discredit the evidence, and the willfulness of the State in failing to disclose the new evidence.” Matthews, 299 Ill. App. 3d at 919.
The undisclosed evidence here has particularly strong effect. If the DNA found on the swab of L.W. does not match L.W’s boyfriend, then that semen came from the man who attacked her. The test of defendant’s DNA shows that he is not the man who produced that semen. The prosecution persuaded the court that the assaults on L.B. and T.S. had enough in common with each other and the attack on L.W. that one man probably committed all three assaults. L.W.’s identification of defendant as her attacker shows that the man who attacked her looked much like defendant, even though a DNA test of L.W.’s boyfriend might conclusively prove that defendant did not attack L.W Thus, the man who assaulted all three victims may be a man who looks like defendant, but who has the DNA most prevalent on the swab from LW
The convictions for the assault on T.S. rested solely on T.S.’s identification of defendant and the evidence that one man committed all three assaults. Prosecutors never requested a test to match the swab of T.S. with defendant. If L.W. mistook defendant for the man who attacked her, T.S. could also have mistaken defendant for the man who attacked her. Even L.B. could have mistaken defendant for the man who assaulted her. The evidence showing a clear match with defendant’s DNA might have resulted from laboratory error, such as contamination of the kind Lewontin described.1 Thus, if the test of *12L.W’s boyfriend confirms the defense expert’s theory, defendant probably should be acquitted for the assault on T.S., and might even be acquitted for the assault on L.B.
The evidence concerning the assaults on T.S. and L.B. may become very closely balanced, depending on the result of the DNA test of L.W’s boyfriend. Prior notice would have permitted the tests necessary to refute the prosecution’s theory that L.W.’s boyfriend, and not the man who assaulted L.W, produced the semen primarily found on the swab of L.W. Due to the prosecution’s knowledge, three years before trial, of the need to look for evidence that L.W had consensual intercourse shortly before the assault, the failure to find the evidence and disclose it to defense in advance of trial appears willful. Accordingly, applying the test restated in Matthews, we hold that the discovery violation requires reversal of the convictions. As defendant does not challenge the sufficiency of the evidence, we remand for a new trial.
II
We address some other issues that may arise again on retrial. Defendant contends that the trial court should have suppressed identifications derived from Officer Johnson’s unreasonable seizure of defendant. Our supreme court, in People v. Murray, 137 Ill. 2d 382, 388-90, 560 N.E.2d 309 (1990), adopted the definition of “seizure” set forth in United States v. Mendenhall, 446 U.S. 544, 64 L. Ed. 2d 497, 100 S. Ct. 1870 (1980).
In Mendenhall drug agents approached the defendant as she walked through an airport. After identifying themselves, the agents asked to see the defendant’s identification and airline ticket. The Court held that no seizure occurred when the defendant voluntarily responded to the requests. The Court explained:
“We adhere to the view that a person is ‘seized’ only when, by means of physical force or a show of authority, his freedom of movement is restrained. Only when such restraint is imposed is there any foundation whatever for invoking constitutional safeguards. The purpose of the Fourth Amendment is not to eliminate all contact between the police and the citizenry, but ‘to prevent arbitrary and oppressive interference by enforcement officials with the privacy and personal security of individuals.’ [Citation.] As long as the person to whom questions are put remains free to disregard the questions and walk away, there has been no intrusion upon that person’s liberty or privacy as would under the Constitution require some particularized and objective justification.
*13We conclude that a person has been ‘seized’ within the meaning of the Fourth Amendment only if, in view of all of the circumstances surrounding the incident, a reasonable person would have believed that he was not free to leave.” Mendenhall, 446 U.S. at 553-54, 64 L. Ed. 2d at 509, 100 S. Ct. at 1877.
Officer Johnson’s conduct here similarly does not qualify as a seizure. When Johnson approached defendant and two other men as they talked on the street, the two men walked away. This provides a strong indication that reasonable persons would have believed they were free to walk away when Johnson approached. The questions Johnson asked defendant are strikingly similar to the questions in Mendenhall, as Johnson asked defendant only for his identification and where he was going. The trial court did not commit manifest error by denying the motion to suppress identifications derived from Johnson’s contact with defendant. See Murray, 137 Ill. 2d at 387.
Next, defendant challenges the admissibility of Sergeant Duffin’s testimony that defendant said he owned a gun. According to Duffin, defendant made the statement in response to Duffin’s question while riding in a police car to a police station and without any prior Miranda warnings.
Our supreme court has explained that police must give Miranda warnings prior to any custodial interrogation. People v. Brown, 136 Ill. 2d 116, 124, 554 N.E.2d 216 (1990).
“The determination of whether an interrogation is custodial should focus on all of the circumstances surrounding the questioning, such as: the location, length, mood and mode of the interrogation; the number of police officers present; any indicia of formal arrest or evidence of restraint; the intentions of the officers; and the extent of knowledge of the officers and the focus of their investigation. [Citation.] The trial court must examine and weigh these factors, along with the credibility of the witnesses. It then must make an objective determination as to what a reasonable man, innocent of any crime, would perceive if he were in defendant’s position.” Brown, 136 Ill. 2d at 124-25.
The question here occurred in a police car, but not all questions asked in police stations or police cars amount to custodial interrogations. Brown, 136 Ill. 2d at 130-31. The question was very brief, and Duffin’s testimony shows no further questioning. The trial court found Duffin’s testimony credible, including the testimony that the question arose in the course of a conversation with defendant. Duffin’s partner heard the question. Apart from the protective pat down, nothing in the encounter between defendant and police gave the appearance of a restraint. The trial court found that a reasonable person in defendant’s position would have believed he was free to leave. Therefore, the court *14found no custodial interrogation, and no need for Miranda warnings. We cannot say that the ruling constitutes manifest error. See Brown, 136 Ill. 2d at 125.
Defendant also asks us to review the judge’s statements, during cross-examination of the prosecution’s experts, concerning the source of the police lab’s DNA database. The prosecution argues that the trial court could take judicial notice of those sources. The trial court may take judicial notice of facts capable of immediate and accurate demonstration by resort to easily accessible sources of indisputable accuracy. In re Marriage of DeBow, 236 Ill. App. 3d 1038, 1040, 602 N.E.2d 984 (1992). The prosecution does not suggest any source for information concerning the police lab’s database, let alone a readily accessible source of indisputable accuracy. The court must refrain from offering testimony for the prosecution from the bench on retrial. See People v. White, 241 Ill. App. 3d 291, 298-99, 608 N.E.2d 1220 (1993).
Finally, defendant contends on this appeal that the court erred by holding the evidence of attacks on L.B. and L.W admissible in the case concerning T.S. While the parties’ positions on this evidence may change on remand, we believe the issue may arise again.
In People v. Berry, 244 Ill. App. 3d 14, 613 N.E.2d 1126 (1991), the court restated the pertinent principles and applied them to conspicuously similar facts. In Berry, a man came up behind the victim late at night, grabbed her and covered her mouth with his hand. He said, “[T]his is a stick-up” and told her if she screamed he would kill her. 244 111. App. 3d at 16. The man shoved the victim into a nearby gangway, took her money, and demanded oral and vaginal sex. Less than three weeks later, shortly after midnight, a man asked a second victim for directions. He then grabbed the victim and put a hand over her mouth and told her if she screamed he would hurt her. He first demanded money, then forced her into a gangway where he ordered her to take off her clothes and masturbate. He then demanded oral and vaginal sex. Each victim described the offender as a muscular black man with curly hair, but the descriptions had little further detail. Both victims identified defendant as the assailant. The trial court permitted testimony concerning the second crime into evidence at a trial about the attack on the first victim. The court accepted the prosecution’s theory that the two attacks showed a single modus operandi.
The appellate court said:
“Generally, evidence of other crimes is inadmissible if relevant merely to establish a defendant’s propensity to commit crime. [Citation.] Evidence of the commission of other crimes is admissible, however, when such evidence is relevant to prove any purpose other *15than propensity to commit crime, such as modus operandi, intent, motive, or absence of mistake. [Citations.] It is within the sound discretion of the trial court to determine whether evidence of other crimes is relevant to a material issue and whether the probative value of such evidence outweighs its prejudicial impact; this determination will be overturned only if there exists a clear abuse of discretion. [Citation.]
The State proffered the other crimes evidence in this case pursuant to the modus operandi exception. The modus operandi or ‘method of working’ exception refers to a pattern of criminal behavior so distinct that separate offenses are recognized as the work of the same person. [Citation.] Between the offense offered to prove modus operandi and the offense charged, there must be a clear connection which creates a logical inference that, if defendant committed the former offense, he also committed the latter. [Citation.] This inference arises when both crimes share peculiar and distinctive features not shared by most offenses of the same type and which, therefore, earmark the offenses as one person’s handiwork. [Citation.] The offenses need not he identical but must share features which, although common to similar crimes in general, are distinctive when considered together.” Berry, 244 Ill. App. 3d at 21.
The appellate court in Berry noted the similarities in time, location, and some of the assailant’s words and actions, but the court agreed with the defendant that no feature of the crimes could be considered unique. The court also noted the significant differences in methods and demands, but concluded that the circuit court did not clearly abuse its discretion by allowing testimony concerning the second crime into evidence. Berry, 244 Ill. App. 3d at 22; see also People v. Johnson, 97 Ill. App. 3d 1055, 1067-68, 423 N.E.2d 1206 (1981).
Here, too, we find the similarities in location, time, and method fail to show any unique features, and the crimes have some dissimilarities. Nonetheless, we find no abuse of discretion in the decision to admit the evidence of the crimes against L.B. and L.W into evidence in the case concerning the assault on T.S. On remand, the court should allow into evidence testimony concerning all three crimes in trial on charges related to the attacks on L.B. and T.S., especially if the DNA of L.W’s boyfriend proves that she misidentified defendant as the man who attacked her. See People v. Tate, 87 Ill. 2d 134, 141-43, 429 N.E.2d 470 (1981) (court employed same standard to determine admissibility of other crimes evidence offered by defense as applied to prosecution’s evidence of other crimes).
The evidence sufficiently supports the trial court’s conclusion that *16Officer Johnson did not seize defendant when he spoke to him briefly on the street. Sergeant Duffin’s testimony also supports the decision to admit defendant’s statement to Duffin into evidence. The court properly admitted testimony about all three crimes into evidence at the trial concerning each individual crime because the crimes sufficiently showed a single modus operandi. But the prosecution’s failure to disclose the identity of a possible source of the semen found on the swab from L.W requires reversal and remand for a new trial on the charges related to the assaults on L.B. and T.S., particularly because timely disclosure would have given the defense an opportunity to develop exculpatory evidence.
Ill
On petition for rehearing the prosecution argues primarily that “the State did not have to make defendant’s case for him.” The prosecution apparently misunderstands its duties. A prosecuting attorney
“is the representative not of an ordinary party to a controversy, but of a sovereignty whose obligation to govern impartially is as compelling as its obligation to govern at all; and whose interest, therefore, in a criminal prosecution is not that it shall win a case, but that justice shall be done. As such, he is in a peculiar and very definite sense the servant of the law, the twofold aim of which is that guilt shall not escape or innocence suffer.” Berger v. United States, 295 U.S. 78, 88, 79 L. Ed. 1314, 1321, 55 S. Ct. 629, 633 (1935).
Our supreme court emphasized:
“It is the duty of the prosecuting officers, as representatives of all the People of the State, not merely to endeavor to obtain convictions in criminal cases but, if the evidence so warrants, to see that a person unjustly accused of crime is absolved from the stigma of a conviction and discharged from custody.” People v. Schoos, 399 Ill. 527, 532, 78 N.E.2d 245 (1948), overruled on other grounds, People v. Izzo, 14 Ill. 2d 203, 209, 151 N.E.2d 329 (1958).
While prosecuting attorneys ordinarily rely on police and other agencies for investigation of criminal acts, the attorney has an affirmative duty to investigate the facts to determine whether the accused has committed an offense. People v. Nohren, 283 Ill. App. 3d 753, 758, 670 N.E.2d 1208 (1996). Thus, when the prosecutor finds evidence casting doubt on the guilt of the accused, the prosecutor has a duty to pursue that evidence to determine whether the State’s Attorney should continue to prosecute the accused.
Here, the DNA test of the swab from L.W. showed a principal banding pattern matching neither L.W. nor defendant. The prosecut*17ing attorney knew then that the pattern derived from either L.W.’s consensual sexual partner, contamination, or the person who actually raped L.W. The prosecutor relied on the theory that the principal banding came from consensual intercourse that took place several days before the rape, and the rapist left very little DNA trace of the rape. The rapist might have left a faint pattern, the prosecution suggests, due to the rapist’s “low sperm count.” But the strong banding pattern found from the test of the swab taken from L.B. showed that defendant did not suffer from such low sperm count — unless the test of the swab from L.B. reflected contamination by defendant’s blood. Therefore, the prosecution had evidence giving considerable reason to doubt that defendant actually raped L.W.
A reasonably straightforward investigation could help resolve the doubt. The prosecutor only needed to contact its witness, L.W., to ask her, as Fish recommended, with whom she had consensual intercourse before the rape. If that man agreed to give blood, the laboratory could determine whether he provided the sperm most evident on the swab of LW
The prosecutor’s obligation to perform this investigation goes beyond the obligation not to prosecute the defendant for a crime he did not commit. If the lab test shows that L.W’s boyfriend does not match the principal banding pattern on the swab, then the prosecutor would have very strong evidence that a rapist, with the DNA found on the swab, remained at large, unprosecuted for his offense. The prosecutor fails to meet the duty to see that the guilty do not escape if he fails to look for the person who raped L.W.
The puzzling decision not to test the swab from T.S. takes on a more ominous appearance when viewed in light of the prosecution’s miscontruction of official duties. If the prosecutor decided not to test the swab because doing so might “make defendant’s case for him,” then the prosecutor has pursued a prosecution despite substantial doubt of the defendant’s guilt. When the prosecutor has reason to doubt the guilt of the accused, he must pursue the evidence to determine whether to continue the prosecution.
The State points out that the public defender also failed to pursue all possible evidence concerning the identity of the person or persons who raped L.B., T.S. and L.W. The duties of other governmental officers, and the arguable neglect of such duties, have no effect on the prosecutor’s duties.
The prosecutor has no obligation to clear cases. He has a duty to use all lawful means that respect the constitutional rights of individuals to investigate illegal activity and discover the perpetrators of such activity. When the prosecutor finds evidence that a defendant may not *18have committed the crime for which he stands accused, the prosecutor has an affirmative duty to pursue that evidence and determine the identities of the actual perpetrators, as part of his duty to see that the innocent should not suffer punishment. Because the petition for rehearing demonstrates misunderstanding of these duties, the petition is denied.
Reversed and remanded.
TULLY, J., concurs.