Opinion
RUSHING, P. J.Defendant Cameron Lee Earle was charged in two separate cases with indecent exposure, a misdemeanor, and sexual assault, a felony. The charges arose from entirely distinct and dissimilar incidents with no apparent historical connection to one another. After first ordering the charges consolidated, the trial court denied a motion by defendant to sever them for trial.
At trial defendant tacitly conceded the indecent exposure charge, proof of which was strong. The underlying conduct occurred in broad daylight; the perpetrator’s license number, which the victim recorded, belonged to a car then owned by defendant; and the victim positively identified defendant from a properly conducted photographic lineup.
The assault case was considerably weaker. The underlying incident occurred at night in a parked car illuminated only by overhead parking lot lights. The victim’s description of her assailant and his vehicle did not match defendant or his vehicle. She characterized her assailant as looking Mexican, with light brown skin resembling her own; but defendant presents a distinctly pallid, European appearance in photographic exhibits, was described by a police booking officer as “white,” and apparently looked European to the *379victim of the indecent exposure, whose description to police resulted in a photographic lineup all but one of whose subjects has an unmistakably northern European appearance. The victim described her assailant as “skinny,” but photographs of defendant suggest an athletic build. The victim failed to notice anything unusual about the assailant’s forehead or ears, but the photographs depict a deeply furrowed brow and protruding, possibly damaged ears. The pictures show an unmistakably athletic bull neck that, like defendant’s prominent ears and deeply furrowed brow, contrasts distinctly with the assailant’s features in a police sketch.
The victim described the assailant’s vehicle as a 1986 Ford Bronco, but the only other witness said it was a pickup. The victim seemed to confirm her own original perception shortly after the assault, when she reported a Ford Bronco to police as possibly the suspect’s vehicle. But defendant did not drive a 1986 Bronco; he drove a 1981 Chevrolet Blazer. Furthermore, defendant was a world-class competitor in the sport of “submission grappling,” but the victim of the assault managed to break the assailant’s grasp, escape the vehicle in which he sought to subdue her, and flee the scene.
These and other facts, discussed in greater detail below, provided fertile ground for a reasonable doubt in jurors’ minds that the victim had correctly identified defendant as her assailant. Given this background we conclude that the court committed reversible error by permitting the prosecution, through the expedient of a joint trial, to place the strongly incriminating evidence of the misdemeanor charge before the same jury that would have to decide the much more difficult felony assault charge—a charge to which it was irrelevant, at least in the absence of foundational evidence that was not presented. This led to a grossly unfair trial in which the prosecutor explicitly urged the jury to convict defendant of the assault based upon his commission of the indecent exposure, which the prosecutor compared to DNA evidence and “modus operandi,” and cited as proof that defendant was a “predator” and “scary guy.” Indeed the prosecutor relied on a variety of spurious legal theories to place the indecent exposure before the jury, as evidence of the assault, as often and emphatically as possible. The indecent exposure thus played a central role, and quite possibly a decisive one, in securing a conviction on the assault charge. We have thus concluded that the trial court abused its discretion by denying the motion for separate trials, and that even if that ruling was within its discretion, the resulting trial was so grossly unfair as to deny defendant’s right to due process of law. We will therefore reverse the assault conviction.
*380Background
A. The Indecent Exposure
Gina Doe1 testified that around 4:00 in the afternoon of September 30, 2004, while waiting into the San Jose mobilehome park where she lived, she noticed a white Ford sedan driving next to her with its windows down. The driver, whom she identified at trial as defendant, told her to “[c]ome here.” On approaching the car, she saw that defendant was naked from the waist down and masturbating his erect penis. She fled in search of aid and managed to write down defendant’s license number while he turned his car around. As she continued calling for assistance, he drove away.
Gina gave police the license number she had recorded. It was stipulated that this number was registered to a white Ford Probe owned by defendant. Gina subsequently viewed a photographic lineup in which she identified an October 2002 photo of defendant as the man who exposed himself to her. The defense offered no challenge to her testimony.
B. The Assault
Gloria Roe (see fn. 1, ante) testified through an interpreter that in late 2004 she was employed selling tamales from a table in a parting lot outside a San Jose market. On December 30 of that year, she arrived at the lot shortly after 5:00 a.m. to set up her table. At some point she decided to adjust the position of her car. As she entered it, with one leg still outside, a man appeared at the door. She asked him to help her. Instead of replying, he lifted and half-pushed, half-threw her into the front seat, where they began to struggle. He pulled her head back by the hair while she sat on the center divider holding the steering wheel to keep him from forcing her into a lying position.
She asked him if he wanted money. He said no, told her to stop struggling, and said that he had a gun and a knife.2 He spoke to her in what Gloria, who *381does not speak English, described as “kind of broken half Spanish.” Her told her “to allow myself—allow myself to have it done ...” Instead she “kept struggling. I wasn’t just going to let him do what he wanted.” He seemed to become more aggressive and to be infuriated by her apron and “fairly tight pants . . . .” She lifted a leg and tried to push him out of the car. At some point she got him to stop pulling her hair, and was able to open the passenger door. She “pushed him as hard as [she] could and . . . shot out the door on the other side.” She ran to a nearby bakery, where she enlisted the aid of one of its workers. She testified that the two of them ran back to the scene of the assault in time to see the assailant drive away.
C. Description of, and Opportunity to Observe, Assailant
Gloria testified that the sun had not yet risen when she was attacked, and there was no light on inside her car. However, she testified, the driver’s door was open and she could see the assailant’s face from the overhead parking lot lights. She insisted that she got a good look at his face from a distance of about six inches. Interviewed by Detective Jorge Gutierrez shortly after the attack, Gloria described her assailant as about five feet nine inches tall, and “thin.” She said he was wearing a black jacket.* *3 According to Detective Gutierrez, Gloria described her assailant as “a light-skinned Hispanic male . . . .” However Gloria herself testified not that his skin was “light,” but that it was “light brown,” “similar” to her own. She denied using the term “Hispanic” to police, but acknowledged telling them that her assailant looked “Mexican American.” She testified that she called him “a Pocho,” meaning, “someone who was bom here and has Mexican parents.” She considered him “Mexican-American as opposed to just Mexican,” not because of his appearance but “because of the way he was speaking Spanish, that he didn’t speak it well.”
About a week after the assault Gloria described the assailant to a police artist. She told the artist the resulting sketch, which was introduced into evidence, resembled the assailant. Defense counsel in her summation described the sketch as “look[ing] similar to a photo that was taken of Mr. Earle in . . . 2002.” Absent from the sketch, however, were the receding hairline, deeply furrowed brow, heavy neck, and protruding ears depicted in photographs of defendant. The face in the sketch is also thinner and has a more angular chin than the squarish face shown in photographs of defendant.
*382D. Assailant’s Vehicle
After returning to the area of the assault in the company of the bakery employee, Gloria saw her assailant climb into a truck and drive away. She initially described the truck as “passing] right in front of us,” but on cross-examination said it passed “at a great distance.” She told police it was a “1986 black Ford Bronco.” However, the bakery employee testified that it was a dark “pickup truck.”
Asked at trial whether she remembered “anything unusual” about the truck, Gloria testified, “We noticed that it did not have the back window.” She did not know whether the window was broken out or rolled down, adding, “It was far away that I saw this.” There is no evidence that she mentioned a missing window in her original statement to police; she apparently first mentioned it after seeing defendant’s tmck.
On January 5, 2005—less than a week after the assault, and not far from its location—Gloria saw a vehicle that she thought resembled the assailant’s. It was a black Ford Bronco with a rolled up, intact rear window. She wrote down the license number and phoned Detective Gutierrez, telling him that she thought she had found the assailant’s vehicle. She later learned that the vehicle she saw was not involved in the assault.
It was stipulated that “[f]rom December 30, 2004, through January 18, 2005,” defendant owned “a black 1981 Chevy Blazer.” It apparently had a cracked or missing rear window. At trial Gloria identified it from photographs as the truck she saw driving away from the scene of the assault. Up until she saw it, she had never described the assailant’s vehicle to police as anything but a 1986 Ford Bronco. Asked to explain this discrepancy, she testified that she never meant the year as anything more than an estimate. She also testified, “I didn’t get that close of a look to see the exact make of the car. And I don’t really know the types of truck that well, but to me it looked like the type of car—it looked to me like a Bronco. And then I found out later that it was actually a Chevy that looks like a Bronco.”4 A friend of hers in Mexico had a Bronco, and she herself had a sport utility vehicle, a 2003 GMC Envoy.
*383E. Defendant’s Identification and Arrest
In the early afternoon of January 18, 2005, Gloria drove her own truck to a carwash about 10 minutes from the location of the assault. While seated in a waiting area looking out through some windows, she saw “from afar” what she described as “the same truck and the same guy” who had attacked her. He was sitting in his truck in the parking lot of a nearby store, facing in her direction. She watched him for a couple of minutes, she testified, “to see if that was the person that had attacked [me].” She acknowledged that she could not see his eyes, height, weight,, or whether he was Hispanic. Nor could she read the truck’s license plate. Asked if she could tell at that distance whether it was a Blazer or a Bronco, she replied, “All I know is that it was the person who attacked me and it was the vehicle.”
When she was “really sure ... it was . . . him”—still before she was close enough to read the license plate—she telephoned Detective Gutierrez and left him a message. She continued to watch the man in the truck for several minutes as he sat “looking all around, kind of like he was looking for someone.” He then moved over to another parking space. “He just kept moving around from one parking space to another. He would be in one space for a few minutes and then would move to another space.”
On the telephoned advice of her brother, she left the carwash to get the license number. She followed the truck on foot as it moved into another store’s parking lot. When she saw it coming back towards her, she ducked into a store, writing down the license number as it passed. Her brother arrived and picked her up. They continued to watch the truck as it moved from parking space to parking space. At some point Gloria’s own truck emerged from the carwash and the black truck left the area.5 By now Gloria was in touch with the police, who told her to follow him. She and her brother complied.
The driver made no attempt to evade them, though they were directly behind him. After driving a few blocks he stopped, got out, and went into a house. Two or three minutes later, the police arrived and went to the door. Shortly thereafter they escorted defendant out of the house, exhibited him to Gloria, and asked if this was the man who had attacked her. At that moment *384he was surrounded by uniformed police officers. She said “yes, that he was the person.” He was wearing “the same black jacket that he had been wearing on the day he attacked [her].” (See fn. 3, ante.) He also had something on his head. Her assailant had not worn a hat, and the man she saw at the carwash had not worn a hat.
F. Charges and Proceedings
Defendant was charged by information with assault accompanied by intent to commit rape (Pen. Code, § 220), assault using a deadly weapon (Pen. Code, § 245, subd. (a)(1)), and indecent exposure (Pen. Code, § 314, subd. 1 (section 314(1))).6 The trial court denied defendant’s motion to try the assault separately from the indecent exposure charges. Before the jury, no defense was offered to the indecent exposure charges, but the assault charges were vigorously contested on the ground that Gloria’s identification of defendant and his truck was mistaken. In support of that theory, the defense elicited testimony from an expert on eyewitness identifications, who was limited on prosecution motion to relevant general principles of perception and recollection. Jurors deliberated for five and one-half hours over two days. During deliberations they sought a readback of testimony concerning the language in which the assailant spoke. Ultimately the jury found defendant guilty of all charges and sustained an allegation that he had personally used a deadly weapon in the assault.
After trial, defendant failed to appear for a probation interview and then for a scheduled court hearing. He apparently remained at large for about five months. Upon his return to custody he was apparently charged with, and eventually entered a guilty plea to, willful failure to appear (Pen. Code, § 1320.5). The trial court sentenced him to consecutive terms of four years for assault with intent to commit rape, one year for personal use of a deadly weapon, and eight months for failure to appear, for an aggregate term of five years eight months.7
Defendant filed this timely appeal.
*385Discussion
I. Order Denying Separate Trials
In his motion to try the indecent exposure charge separately from the assault charge, defendant contended that the latter was relatively weak in that it depended on the credibility of the victim’s identification of defendant and his vehicle.8 Defense counsel contended in written argument that joint trials would be prejudicial because evidence of the indecent exposure, which would be inadmissible in a separate trial of the assault, would tend to make the jury think that he had committed the latter offense in part through improper inferences about criminal character. Defense counsel contended that the indecent exposure was “inflammatory” in the context of the assault charge, and constituted a strong case, which the prosecutor would use to bolster the weaker assault case by “aggregating] ” the evidence. The prosecutor did not submit written opposition, but argued at the hearing, among other things, that the two charges were admissible against each other and that the assault case was not weak in light of the victim’s identification and the police artist’s sketch, which he described as “incredibly similar, almost identical to a photograph of the defendant that was used in the photographic lineup . . . .”
In support of the motion, defense counsel submitted a declaration by Brian Abbott, Ph.D., a clinical and forensic psychologist with 28 years’ experience in treating and evaluating sex offenders. Dr. Abbott declared that roughly 20 to 30 percent of criminal exhibitionists may also commit rape.9 He drew a distinction between “primary exhibitionism,” in which the actor finds exposing himself to be sexually gratifying in itself, and “transitional exhibitionism,” in which “acts of indecent exposure serve as a psychological step toward engaging in hands-on sexually assaultive behavior.” In the latter situation, the subject fantasizes about rape or other assaultive sexual behavior while exposing himself, and “[o]ver time, the repeated fantasizing about rape serves to breakdown [íz'c] the exhibitionist’s inhibitions against acting in such ways.” Because this progression is subject to various “social, physiological, and personality dynamics,” it is impossible to predict which exhibitionists are likely to, or will, commit more serious offenses. In sum, “Exhibitionistic behavior alone cannot be considered indicative of or a precursor to an individual committing an act of rape. In fact, it is more likely exhibitionists will only engage in exhibitionistic behavior rather than to progress to acts of hands-on sexual assault, such as rape.” Accordingly, “it would be inappropriate and prejudicial for a trier of fact to hear evidence that a person who *386commits one or more acts of exhibitionism would be reasonably likely to commit rape in the future.”10
The court denied the motion for separate trials. It explained its ruling only by saying, “the defendant has not shown that he’s prejudiced by both charges being tried together . . . .”
II. Governing Principles
Penal Code section 954 (section 954) addresses, among other things, the joinder of criminal charges in pleadings and their severance for trial. As relevant here it provides: “An accusatory pleading may charge two or more different offenses connected together in their commission, or different statements of the same offense or two or more different offenses of the same class of crimes or offenses, under separate counts .... [T]he court in which a case is triable, in the interests of justice and for good cause shown, may in its discretion order that the different offenses or counts set forth in the accusatory pleading be tried separately . . . .” (§ 954.)
The first of these clauses permits the prosecutor to combine multiple charges in a single accusatory pleading provided they meet the stated criteria, i.e., are “connected together in their commission, or [constitute] different statements of the same offense[,] or . . . [belong to] the same class of crimes or offenses . . . .” The second clause assumes that the charges have already been joined in the pleadings but empowers the court, “in its discretion,” to order them tried separately “in the interests of justice and for good cause shown . . . .” (§ 954.) These provisions raise distinct questions: Whether the charges are eligible for joinder, i.e., satisfy the statutory criteria for joint pleading; and whether, despite their eligibility, they ought to be tried separately to ensure a fair trial. (See People v. Hill (1995) 34 Cal.App.4th 727, 734 [41 Cal.Rptr.2d 39] [“[E]ven where joinder is statutorily authorized, severance may be required if joinder results in prejudice so great as to deny the defendant a fair trial.”].) Defendant contends that both of these questions point to error, i.e., that the indecent exposure charge was ineligible for joinder with the assault charge and that, even if the charges were properly joined in the first instance, they could not be fairly tried together. Because we find the latter point dispositive, we do not address the former. 11
*387The statute explicitly vests the trial court with discretion to determine whether to order separate trials. (§ 954.) This of course triggers the deferential abuse-of-discretion standard of appellate review. To successfully challenge the denial of a severance motion, an appellant must “make a ‘ “clear showing of prejudice” ’ ” and that the ruling “ ‘ ‘ “ ‘falls outside the bounds of reason,’ ” ’ ” ’ ” (People v. Soper (2009) 45 Cal.4th 759, 774 [89 Cal.Rptr.3d 188, 200 P.3d 816] (Soper), italics omitted, quoting Alcala v. Superior Court (2008) 43 Cal.4th 1205, 1220 [78 Cal.Rptr.3d 272, 185 P.3d 708].) This demands a stronger showing than would be sufficient to require reversal based on the erroneous admission of evidence of uncharged crimes. (Soper, at p. 774.)
The trial court’s denial of a motion to sever must be evaluated in light of the facts and circumstances apparent to the court at the time of its ruling. (Soper, supra, 45 Cal.4th at pp. 774—776, fn. 10.) However, “ ‘[e]ven if a trial court’s severance or joinder ruling is correct at the time it was made, a reviewing court must reverse the judgment if the “defendant shows that joinder actually resulted in ‘gross unfairness’ amounting to a denial of due process.” ’ [Citation.]” (People v. Macklem (2007) 149 Cal.App.4th 674, 698 [57 Cal.Rptr.3d 237]; accord, Soper, supra, 45 Cal.4th at pp. 782-783.)
As indicated by the foregoing, the key inquiry before the trial court on a motion to sever is whether joint trials pose an unacceptable risk of prejudice, i.e., of unfairly affecting the adjudication of one or more of the charges. (See People v. Smith (2007) 40 Cal.4th 483, 510 [54 Cal.Rptr.3d 245, 150 P.3d 1224] [“To demonstrate that a denial of severance was reversible error, defendant must ‘ “clearly establish that there [was] a substantial danger of prejudice requiring that the charges be separately tried.” ’ ”].) The chief source of potential prejudice is “spillover effect,” i.e., the risk that evidence not admissible as to one of the charges, but admitted in connection with another, will affect the verdict on the charge as to which it is inadmissible. (See Soper, supra, 45 Cal.4th at pp. 775, 781.) Obviously there can be no spillover effect unless some of the evidence to be heard by the jury is *388inadmissible as to at least one of the charges. Therefore the first consideration in evaluating a motion to sever is “cross-admissibility,” i.e., the extent to which evidence of charge A would have been admissible in a hypothetical separate trial of charge B, and vice versa. (Soper, supra, 45 Cal.4th at pp. 774-775.)
If the jury will be exposed to evidence that is not cross-admissible, then the possibility of spillover effect is present and the court must proceed to evaluate the risk that the jury will be unfairly influenced by it. (Soper, supra, 45 Cal.4th at p. 775, quoting People v. Bean (1988) 46 Cal.3d 919, 938 [251 Cal.Rptr. 467, 760 P.2d 996] [“If we determine that evidence underlying properly joined charges would not be cross-admissible, we proceed to consider ‘whether the benefits of joinder were sufficiently substantial to outweigh the possible “spill-over” effect of the “other-crimes” evidence on the jury in its consideration of the evidence of defendant’s guilt of each set of offenses.’ ”].) Courts have identified several factors to be considered in evaluating the risk. The pertinent ones here are whether the spillover evidence is likely to “ ‘unusually inflame the jury against the defendant,’ ” and whether any of the charges as to which the evidence is not admissible rests on a “ ‘weak case’ ” that may be unfairly bolstered by the spillover evidence. (Alcala v. Superior Court, supra, 43 Cal.4th 1205, 1220-1221, quoting People v. Mendoza (2000) 24 Cal.4th 130, 161 [99 Cal.Rptr.2d 485, 6 P.3d 150]; see Soper, supra, 45 Cal.4th at pp 775, 780.) In essence these are specific articulations of two broader variables: How likely is the spillover evidence to influence the jurors, and how susceptible is the charge to such influence? Once the spillover evidence has been identified and its potential to influence the verdict evaluated, the trial court must weigh the resulting risk of prejudice against the advantages of joint trials. (Soper, supra, 45 Cal.4th at p. 775.)
We now apply these principles to the case at hand.
HI. Cross-admissibility
A. Introduction
We turn to the question of “cross-admissibility,” i.e., the extent to which evidence offered to prove the indecent exposure charge would have been admissible to prove the assault charge, or vice versa. In support of the motion for separate trials, defense counsel asserted that there was no cross-admissible evidence in the case, adding, “[0]bviously, there are no witnesses or evidence that will be used in both charges . . . .” She argued that the indecent exposure and the assault were too dissimilar, too widely separated in time, and too *389unrelated in occurrence for either to have any proper bearing on the other.12 She also asserted that each charge would “rais[e] impermissible innuendoes of character” if introduced in a trial of the other. She expressed the prescient concern that “the People will try to promote that if a person would expose themselves that they are likely or probably would, in fact, commit the violent act of rape,” adding that this would be an “inflammatory” use of the indecent exposure. These statements were sufficient to tender the issue whether, in a separate prosecution of the assault charge, evidence of the indecent exposure would be excluded on grounds that it was irrelevant, that its potential for prejudice outweighed any probative value it might be found to have, or that it violated the general prohibition against using evidence of bad character to prove the commission of a crime. (See Evid. Code, §§ 352 (section 352), 1101 (section 1101), subd. (a).)
The prosecutor contended orally that the assault evidence would be admissible in a separate trial of the indecent exposure, and vice versa, “under [Evidence Code sections] 1108 and 1101[, subdivision] (b).”
The trial court expressed no opinion on cross-admissibility, instead reciting the boilerplate finding that defendant had failed to demonstrate that a joint trial would cause him undue prejudice.
B. Intent
Respondent suggests that the indecent exposure was relevant to the assault in that it could support an inference that the latter was committed with intent to commit rape, an element of one of the two aggravated assault charges. On the evidence before the jury, such an inference would be purely speculative. It necessarily rests on the supposition that where a man has criminally exposed himself on one occasion, and he thereafter commits an assault, it may reasonably be inferred from the exposure incident that he acted in the second incident for the purpose of committing rape rather than, e.g., to secure control of the victim in order to obtain property, or to express irrational rage. The evidence before the jury afforded no basis for such a supposition. For all the jury could know, criminal exhibitionists as a class might be characterized by an unusual inhibition against physical contact, and a particular inability to *390commit sexual assault. They might thus be less likely than randomly selected members of the public to commit such a crime, or form the intent to do so. That this hypothesis is apparently contrary to the only expert opinion in the record, as discussed in more detail below, cannot sustain an inference the jury had no basis to draw.
Further, this theory begs the only real question in the case, which is whether defendant was the assailant. The proposed inference of intent necessarily supposes that he was; its necessary premise is that the indecent exposure and the assault were committed by the same person, i.e., defendant. But jurors were given many quite substantial reasons, discussed in detail below, to doubt this premise. At the same time, they had no reason at all to doubt that the purpose of the assault, whoever committed it, was to rape Gloria. According to her account of the attack, which went entirely unchallenged at trial, the assailant all but told her he intended to rape her. After expressly denying that he wanted money, he told her to stop struggling and to “allow [herself] to have it done . . . .” She described him as furiously stymied by the obstruction of her apron and “fairly tight pants . . . ,”13 Unless the jury simply refused to believe her account—in which case it would have acquitted defendant without worrying about the assailant’s intent—there could be no doubt about the issue. The prosecutor told the jury as much early in his argument: “[T]here’s no question—and . . . nothing to be disputed that the victim was assaulted by a person with the intent to commit rape.” He soon repeated the point: “There’s no question that the perpetrator was assaulting her with the intent to commit rape.” We agree, as would any reasonable juror. Intent simply was not an issue on which a reasonable juror could entertain any doubt—at least, not without doubting the entire prosecution case.
This is not to suggest that a defendant can “limit the prosecution’s evidence” merely by “ ‘not putting things at issue.’ ” (People v. Thornton (2000) 85 Cal.App.4th 44, 48 [101 Cal.Rptr.2d 825].) A mere failure to contest an element of the prosecution’s case cannot by itself bar evidence to *391prove that element. But before potentially prejudicial evidence can be admitted to show an element of the offense there must be some concrete basis to suppose that the jury might fail to find that element beyond a reasonable doubt. (See People v. Balcom (1994) 7 Cal.4th 414, 423 [27 Cal.Rptr.2d 666, 867 P.2d 777] [“[B]ecause the victim’s testimony that defendant placed a gun to her head, if believed, constitutes compelling evidence of defendant’s intent, evidence of defendant’s uncharged similar offenses would be merely cumulative on this issue.” “Accordingly, . . . the limited probative value of the evidence of uncharged offenses, to prove intent, is outweighed by the substantial prejudicial effect of such evidence”]; People v. Thornton, supra, 85 Cal.App.4th at p. 49) This was precisely the case here: even if the indecent exposure had possessed some logical tendency to establish the intent element of the assault charge—which, again, it did not—its probative value on that point was wholly theoretical, and it would have been a patent abuse of discretion to admit evidence of the indecent exposure for this purpose in a separate trial of the assault.
The dissent attempts to save this rationale by exploiting ambiguities in the term “intent.” This approach is encapsulated in the assertion that “[t]he two offenses have very similar intent elements.” (Dis. opn., post, at p. 418, fn. 4.) This is true, if at all, only in some lay sense, and an abstract, impressionistic one at that: paradigmatically, both crimes involve intentional sexual imposition upon, and at least potential insult to another, albeit psychological in one case and physical in the other. But if that gives them “very similar intent elements,” then the same may be said of embezzlement and armed robbery. Indeed, all sex crimes have “very similar intent elements,” and the point the dissent is attempting to refute—that the two crimes here shared nothing but their sexual character—is instead reinforced.
The dissent’s blurring of the concept of “intent” is also reflected in its assertion that the indecent exposure was relevant to intent because it “showed . . . defendant’s sexual intent toward lone female strangers.” (Dis. opn., post, at p. 424.) But as the dissent goes on to acknowledge, the requisite mental state for the assault charge was intent to commit rape, not “lewd sexual intent toward female strangers.” (See dis. opn., post, at p. 424; see People v. Trotter (1984) 160 Cal.App.3d 1217, 1222 [207 Cal.Rptr. 165], quoting People v. Cortez (1970) 13 Cal.App.3d 317, 326 [91 Cal.Rptr. 660] [“Courts recognize ‘a distinction . . . between the intent to rape, and lewdness, indecency and lasciviousness either alone or accompanied by an intent to seduce.’ ”].) The proper use of the term “intent” in the present context is the mental element—the mens rea—that the prosecution is required to prove in order to make a prima facie case of guilt. (People v. Ewoldt (1994) 7 Cal.4th 380, 394, fn. 2 [27 Cal.Rptr.2d 646, 867 P.2d 757] [“Evidence of *392intent is admissible to prove that, if the defendant committed the act alleged, he or she did so with the intent that comprises an element of the charged offense.” (Italics omitted.)].) The mens rea for ordinary assault is a general intent to commit an act that “by its nature would directly and probably result in the application of force to someone . . . .” (CALCRIM No. 875; see People v. Colantuono (1994) 7 Cal.4th 206, 214 [26 Cal.Rptr.2d 908, 865 P.2d 704].) Assault with intent to commit rape further requires, as the name says, the specific intent to commit rape, i.e., sexual intercourse by force. (People v. Trotter, supra, 160 Cal.App.3d at p. 1224; see CALCRIM No. 890.) These elements bear no similarity to the mental elements for indecent exposure, which are that (1) the defendant acted “willfully” (§ 314(1)), meaning essentially that he knew he was exposing his genitals (see In re Smith (1972) 7 Cal.3d 362, 364 [102 Cal.Rptr. 335, 497 P.2d 807] (Smith)); (2) he “intended by his conduct to direct public attention to his genitals” (id. at p. 366); and (3) he acted “for purposes of sexual arousal, gratification, or affront” (ibid., fn. omitted).
The dissent’s attempt to equate these quite distinct mental states may be understood to rest on some conceptual or symbolic parallel between indecent exposure and rape in that a paradigmatic act of indecent exposure— such as the one shown here—involves a kind of assault upon the victim’s sensibilities. Perhaps the dissent means to imply that indecent exposure typically entails the infliction, or intent to inflict, a mental or emotional outrage or insult to the victim, which conceptually resembles the physical insult or outrage inflicted on the victim of a sexual assault. But this kind of symbolic comparison simply will not substitute for a focus on the actual elements of the offenses. Moreover, while insult is a common feature of the paradigmatic indecent exposure, it is not a necessary element of the crime. A purpose of “affront[ing]” the victim may satisfy part of the mens rea, but it is not necessary; the crime is also complete if the defendant acts for purposes of his own or the victim’s (presumably imagined) “arousal” or “gratification,” whether or not he intends to cause “affront” or believes he is doing so. (§ 314(1).)
The indecent exposure was simply not relevant to show the intent required for the assault charge, and could not be admitted for that purpose.
C. Motive
Intermingled with respondent’s argument on the issue of intent is the suggestion that the indecent exposure would have been admissible to show a sexual motive for the assault. Properly understood, the motive for a crime is never an issue in its own right, but may operate as a basis to establish identity on the rationale that the defendant’s possession of a reason (motive) to *393commit the charged offense increases the likelihood that he did so. The concept is familiar to anyone who has read or watched a detective story. One who stands to inherit money from a murder victim will routinely come under suspicion for the death, and if he is charged, the fact of his expectancy may be offered in court as circumstantial evidence that it was indeed he who killed the victim.
Here the attempt to justify the trial court’s ruling on such a theory fails because defendant’s commission of an indecent exposure simply could not show, without more, a motive to commit rape—except, yet again, through the undemonstrated premise that one who commits indecent exposure also wants, needs, or wishes to commit rape. Nor does the dissent appear to defend this theory of admissibility—except by blurring motive and intent, as in the assertion that the “quintessential element” of sexual assault is “the sexual motivation for the assault—the intent to commit rape.” (Dis. opn., post, at p. 418.) But motive is a much broader concept than intent; it is not an element of the offense at all; and it is relevant only insofar as it tends circumstantially to increase the likelihood that defendant, rather than another, committed the charged offense. Here no one questions that both crimes were sexually motivated. But many people other than defendant engage in some conduct for sexual motives. The question is in what conduct is defendant sexually motivated to engage? On this record, evidence of the indecent exposure had no tendency at all to show that he had a motive to commit sexual assault. The proposed inference of motive is no more logical than saying, “Joe manufactured methamphetamine, therefore he had a motive to steal a car.” Without more, such a statement is simply a non sequitur.
D. Identity
Neither respondent nor the dissent makes any attempt to defend the use the prosecutor actually made of the indecent exposure evidence, which was to distract the jury from weaknesses in his own case on the issue of identity. Indeed the prosecutor explicitly cited the indecent exposure as “modus operandi” evidence, an implicit claim that it showed defendant, as opposed to someone else, to have been Gloria’s assailant. He also claimed that the indecent exposure furnished a substitute for “DNA” in tying defendant to the assault.
The evidence was patently irrelevant for any such purpose. Evidence of uncharged misconduct has long been admissible to prove identity when “the uncharged misconduct and the charged offense . . . share common features that are sufficiently distinctive so as to support the inference that the same person committed both acts.” (People v. Ewoldt, supra, 7 Cal.4th at p. 403, italics added; accord, People v. Matson, supra, 13 Cal.3d at p. 40; *394Soper, supra, 45 Cal.4th at p. 776.) But such an inference depends upon the presence of distinctive “marks” that are “shared by the uncharged and charged crimes.” (People v. Thornton (1974) 11 Cal.3d 738, 756 [114 Cal.Rptr. 467, 523 P.2d 267], overruled on other grounds in People v. Flannel (1979) 25 Cal.3d 668, 684, fn. 12 [160 Cal.Rptr. 84, 603 P.2d 1].) “ ‘[T]he inference of identity arises when the marks common to the charged and uncharged offenses, considered singly or in combination, logically operate to set the charged and uncharged offenses apart from other crimes of the same general variety and, in so doing, tend to suggest that the perpetrator of the uncharged offenses was the perpetrator of the charged offenses.’ ” (Thornton, at p. 756, italics added, quoting People v. Haston (1968) 69 Cal.2d 233, 245-246 [70 Cal.Rptr. 419, 444 P.2d 91].) “ ‘The pattern and characteristics of the crimes must be so unusual and distinctive as to be like a signature.’ ” (Ewoldt, supra, 7 Cal.4th 380, 403; Soper, supra, 45 Cal.4th at p. 776.)
Similarly, an inference of identity can be drawn from “a distinctive modus operand!,” but again, to be admissible on this basis, “the evidence must disclose common marks or identifiers, that, considered singly or in combination, support a strong inference that the defendant committed both crimes.” (People v. Bradford (1997) 15 Cal.4th 1229, 1316 [65 Cal.Rptr.2d 145, 939 P.2d 259].) “These common marks must be distinctive rather than ordinary aspects of any such category of crime. They must be sufficiently distinctive that they bear defendant’s unique ‘signature. ’ Reaching a conclusion that offenses are signature crimes requires a comparison of the degree of distinctiveness of shared marks with the common or minimally distinctive aspects of each crime.” (People v. Bean (1988) 46 Cal.3d 919, 937 [251 Cal.Rptr. 467, 760 P.2d 996], italics added.)
Obviously the two incidents here share no “marks” distinguishing them from other indecent exposures or other sexual assaults. Both involved taking advantage of unwilling victims, but this is true of all attempted rapes and many, if not the vast majority of, indecent exposures. Certainly this feature is not so distinctive as to set these two crimes apart. Much the same is true of the prosecutor’s argument to the jury that the crimes had the same perpetrator because the offender in both cases “sought out a woman that was alone while he was using his car.” Elsewhere he described the perpetrator as singling out a victim “using his own vehicle[,] [a]nd when she tried to escape, he followed her.” Gina’s somewhat confusing testimony could be understood to mean that defendant made some attempt to follow her after initially exposing himself, but there is no evidence that the assailant attempted to follow Gloria after she escaped his grasp and fled. This leaves the commonplace facts that the offender in both cases used an automobile— instead of traveling to and from the scene by foot, bicycle, or helicopter—and *395that he chose a victim who was alone rather than one in a position to readily secure aid. The choice of an isolated victim does not remotely approach the level of distinctiveness required to support an inference of identity. “By their very nature,” most sex crimes are “committed in seclusion without third party witnesses or substantial corroborating evidence.” (People v. Falsetta (1999) 21 Cal.4th 903, 915 [89 Cal.Rptr.2d 847, 986 P.2d 182]; see People v. Thomas (1978) 20 Cal.3d 457, 468 [143 Cal.Rptr. 215, 573 P.2d 433] [noting the “circumstances of privacy and seclusion surrounding the commission of most sex offenses”], impliedly overruled on another point in People v. Tassell (1984) 36 Cal.3d 77, 87-89 & fn. 8 [201 Cal.Rptr. 567, 679 P.2d 1], as stated in People v. Alcala (1984) 36 Cal.3d 604, 634, fn. 18 [205 Cal.Rptr. 775, 685 P.2d 1126].) Nor does anything distinctive appear in the mere use of a vehicle, particularly where, as here, it is not the same vehicle or used in the same manner. The indecent exposure was committed from within the perpetrator’s vehicle; he never left it. The assailant, in contrast, parked away from the victim, snuck up on her, and carried out the attack inside her vehicle. The only feature the two crimes really had in common, aside from their sexual character, was that the perpetrator traveled to and from them by car. Given the obvious advantages of a private automobile for this purpose, it is hardly surprising that innumerable cases share the same feature. (See People v. Bean, supra, 46 Cal.3d at p. 937 [several common features, including taking and abandoning victim’s car, lacked sufficient distinctiveness to establish modus operandi].)
The two crimes had no distinguishing characteristics in common.14 They simply were not similar. They did not even involve the same, or notably similar, acts—a mark that may ordinarily be taken for granted in cases involving the use of uncharged crimes to prove identity. It is one thing to find a man guilty of armed robbery based on his previous commission of that offense. But to base such a finding on an act of shoplifting passes beyond speculation into the realm of hunch or fancy. Here the jury had no rational basis to infer from the facts surrounding the indecent exposure that it was defendant and not someone else who committed the far more serious, violent offense of sexual assault, on another occasion, under entirely different *396circumstances. The indecent exposure was simply not relevant to any issue in the assault case and could not have been introduced in evidence in a separate trial of that case.
E. Propensity
The only colorable basis for admitting evidence of the indecent exposure in a hypothetical separate trial of the assault would have been that it was admissible under Evidence Code section 1108 to show that defendant had a propensity or predisposition to commit sexual assault. The prosecutor alluded to this theory, respondent defends it, and the dissent endorses it in passing. In our view it presents two distinct questions: whether Evidence Code section 1108’s exception to the rule against propensity evidence extends to proof of a wholly different crime; and whether, assuming it does, evidence of defendant’s indecent exposure had any tendency in reason, on the evidence before the jury, to establish such a predisposition. We do not finally reach the first question, but we answer the second with a resounding negative.
It is of course the general rule that evidence of uncharged misconduct is not admissible to prove a predisposition to such conduct, even when it would otherwise be relevant to establish misconduct on a specific occasion-such as the occasion of a charged criminal offense. (See § 1101, subd. (a).) Although the rule is codified in rather indirect terms, its gist is that parties in both civil and criminal cases are generally prohibited from introducing evidence of a general character for good or bad conduct, or a predisposition or propensity to engage in conduct of a particular kind, to show conduct on a particular occasion. (See ibid.) Evidence Code section 1108 creates an exception to this rule, though again the point is expressed rather obliquely: “In a criminal action in which the defendant is accused of a sexual offense, evidence of the defendant’s commission of another sexual offense or offenses is not made inadmissible by Section 1101, if the evidence is not inadmissible pursuant to Section 352.” (Evid. Code, § 1108, subd. (a).)
According to its sponsor, the statute was intended “to provide explicitly that evidence of other offenses within the scope of the section is not subject to § 1101’s prohibition of evidence of character or disposition,” and thereby to “permit[] courts to admit such evidence on a common sense basis— without a precondition of finding a ‘non-character’ purpose for which it is relevant . . . .” (Assemblyman Rogan, letter regarding Assem. Bill No. 882 (1995-1996 Reg. Sess.) published in 2 Assem. J. (1995-1996 Reg. Sess.) p. 3278, reprinted at 29B pt. 3 West’s Ann. Evid. Code (1999 pocket supp.) foil. § 1108, at pp. 40-41 [bill was “modeled on Rules 413-15 of the Federal *397Rules of Evidence, and adapts the principle of these rules to the framework of California law”].) The object is to allow “rational assessment by juries of evidence so admitted. This includes consideration of the other sexual offenses as evidence of the defendant’s disposition to commit such crimes, and for its [j/c] bearing on the probability or improbability that the defendant has been falsely or mistakenly accused of such an offense.” (Ibid; see People v. Reliford (2003) 29 Cal.4th 1007, 1009 [130 Cal.Rptr.2d 254, 62 P.3d 601] [Evid. Code, § 1108 “allows evidence of the defendant’s uncharged sex crimes to be introduced in a sex offense prosecution to demonstrate the defendant’s disposition to commit such crimes” (italics added)].)
The statute “implicitly abrogates prior decisions . . . indicating that ‘propensity’ evidence is per se unduly prejudicial to the defense.” (People v. Falsetta, supra, 21 Cal.4th 903, 911; see People v. Abilez (2007) 41 Cal.4th 472, 502 [61 Cal.Rptr.3d 526, 161 P.3d 58] (Abilez).) Its purpose, however, was to relax the traditional limits, not abolish them. Thus a trial court may not “ ‘admit or exclude every sex offense a defendant commits,’ ” but must “ ‘consider [other] factors’ ” bearing on the relevance, probative value, and prejudicial potential of the evidence, including “ ‘its similarity to the charged offense.’ ” (Abilez, supra, 41 Cal.4th at p. 502, quoting Falsetto, supra, 21 Cal.4th at p. 917, italics added in Abilez.) Obviously, the “lack of similarity” between charged and uncharged offenses can be enough by itself to justify an exclusion of the latter in an exercise of the trial court’s discretion. (Abilez, supra, 41 Cal.4th at p. 502.) Logically, it can also be enough to compel its exclusion where, as here, any inference of predisposition to commit the charged offense would be wholly speculative, i.e., where the uncharged offense has no tendency in reason to show that the defendant actually has the propensity whose proof the statute authorizes.
The statute would clearly authorize the admission of evidence of an indecent exposure in a second prosecution for indecent exposure, on the rationale that the defendant’s commission of the first crime supports an inference that he is predisposed to such conduct, and that since it shared the same essential nature as the conduct underlying the charged offense, its occurrence increased the likelihood that the defendant committed that offense. However, the statute cannot infuse an uncharged offense with relevance or probative value it cannot rationally be found to possess. In order for evidence of another crime to be relevant under Evidence Code section 1108, it must have some tendency in reason to show that the defendant is predisposed to engage in conduct of the type charged. (See People v. Soto (1998) 64 Cal.App.4th 966, 989 [75 Cal.Rptr.2d 605] [“ ‘A defendant with a propensity to commit acts similar to the charged crime is more likely to have committed the charged crime than another.’ ” (Italics added.)].) If the uncharged crime *398does not rationally support such an inference, then it is simply irrelevant in a prosecution for the charged one.
Does the commission of indecent exposure rationally support an inference that the perpetrator has a propensity or predisposition to commit rape? Not without some kind of expert testimony, it does not.15 “[Propensity” refers to “[a] predisposition or inclination to, towards, or for a particular action, habit, quality, etc.; a tendency to do something.” (Oxford English Diet, (draft rev. Sep. 2008) <http://dictionary.oed.com/cgi/entry/50190128> [as of Mar. 19, 2009], italics added.) Similarly, “predisposition” is “[t]he condition of being predisposed or inclined beforehand (to do something, or to a particular opinion, course of action, etc.); a prior inclination or pre-existing tendency.” (Id., <http://dictionary.oed.com/cgi/entry/50186809> [as of Mar. 19, 2009], italics added.) As the use of “tendency” here reflects, both terms refer to a constant, or at least recurring, inclination. (See Oxford English Diet. (2d ed. 1989) <http://dictionary.oed.com/cgi/entry/50248869> [as of Mar. 19, 2009] [defining “tendency” as “[t]he fact or quality of tending to something; a constant disposition to move or act in some direction or toward some point, end, or purpose; leaning, inclination, bias, or bent toward some object, effect, or result” (italics added)].)
There is of course substantial basis to suppose that many sexual offenses arise from a persistent or recurring compulsion, desire, appetite, or drive on the defendant’s part to engage in such conduct. Such a drive constitutes propensity of a kind, and probably a forcefulness, wholly unlike the psychological processes actuating most criminal offenses.16 It is probably safe to say that most people have not the slightest inclination to lewdly expose their *399genitals to strangers, no matter how perfect an opportunity might arise. Those who do engage in such conduct, at least in the paradigmatic manner shown here, typically act not from any circumstantial or situational motive, but from a persistent and recurring desire to exhibit their sex organs. Because the desire is typically persistent and recurring, it is a good bet that someone who commits this act once will be predisposed to commit it again (and again). This provides a solid foundation for a focused inference that a defendant, having criminally exposed himself on one occasion, will do so again. This in turn adds weight to an accusation that he has done so again.
But a propensity to commit one kind of sex act cannot be supposed, without further evidentiary foundation, to demonstrate a propensity to commit a different act. The psychological manuals are full of paraphilias, from clothing fetishes to self-mutilation, some of which are criminal, some of which are not. No layperson can do more than guess at the extent, if any, to which a person predisposed to one kind of deviant sexual conduct may be predisposed to another kind of deviant sexual conduct, criminal or otherwise. Is one who commits an act of necrophilia (Health & Saf. Code, § 7052) more likely than a randomly selected person to commit an act of rape? Child molestation? Indecent exposure? Is a pedophile more likely than a rapist or a member of the public to commit necrophilia? Without some evidence on the subject, a jury cannot answer these questions.
Certainly these questions cannot rationally be answered, as the dissent would answer them, by drawing abstract conceptual analogies between one kind of sexual misconduct and another. That they all involve some kind of outrage against an unconsenting person (or, in the one case, a former person) can hardly justify any particular inference about the likelihood that one such activity shows a predilection for another. Most crimes can be conceptualized as outrages or impositions against one or more unconsenting persons. That does not make it rational to suppose that the perpetrator of one crime will commit another crime of a substantially different character. If anything, such speculation may be even more unwarranted in the case of sexual misconduct, which may reflect highly particularized, compulsive activity rather than opportunistic, situational, or impulsive conduct. Here the dissent would impliedly justify an inference of predisposition by equating the psychic distress inflicted by a criminal exhibitionist to the physical and potentially lethal violence inflicted by a rapist. That is not a reasoned inference; it is sheer guesswork.
*400This is not to assert as a fact that criminal exhibitionists are no more likely than other members of the public to commit sexual assault. The expert declaration submitted by the defense—to the court, not the jury—to establish the prejudicial effect of trying the two charges together showed a correlation between the two types of misconduct. But it is a weak one: according to the studies there summarized, nearly three out of four indecent exposers will not commit rape, and at least three out of four rapists have not committed indecent exposure. Whether that correlation would be sufficient to permit introduction of this evidence on a proper foundation is a question we do not and need not reach, for the prosecution made no attempt to lay any such foundation. The jury therefore had no rational basis on which to draw even the weak inference this evidence might sustain.
On the evidence actually before the jury, defendant’s commission of indecent exposure was simply irrelevant to the assault charge, i.e., it had no tendency in reason to show that he committed the latter offense. As we have said, Evidence Code section 1108 does not purport to make irrelevant evidence relevant. It therefore furnished no justification to admit the indecent exposure in a separate trial of the assault.
IV. Potential Spillover Effect
A. Inflammatory Tendency
Having concluded that evidence of the indecent exposure was spillover evidence as to the assault charge, we must assess its potential to prejudicially affect the jury’s assessment of that charge. (Soper, supra, 45 Cal.4th at p. 775.) In noncapital cases, this inquiry typically revolves around two factors: the extent to which the spillover evidence was particularly “ ‘likely to unusually inflame the jury against the defendant,’ ” and the extent to which “ ‘a weak case [was] joined with a strong case ... so that the total evidence m[ight] alter the outcome of some or all of the charges . . . .’ ” (Alcala v. Superior Court, supra, 43 Cal.4th 1205, 1220-1221; see Soper, supra, 45 Cal.4th at p. 775.)
The presence of the first factor here is debatable. Surely in many contexts an act of indecent exposure would not be expected to “inflame” a jury’s sentiments, however obnoxious they might find it. Notwithstanding its loathsomeness, it remains a distinctly nonviolent offense. (See In re Lynch (1972) 8 Cal.3d 410, 429-431 [105 Cal.Rptr. 217, 503 P.2d 921] (Lynch).) While its commission often appears “ ‘intended to evoke fear and shock’ ” (id. at p. 430, quoting Gigeroff et al., Sex Offenders on Probation: The Exhibitionist (1968) 32 Fed.Prob. (No. 3) 17, 19), no such effect is necessary to its completion. So long as the defendant acts with the requisite intent, the crime *401requires only that the exposure occur in a “public place” or in the presence of “other persons to be offended or annoyed thereby . . . .” (§ 314(1).) The second phrase is satisfied so long as “there are persons present who may . . . be ‘offended or annoyed’ . . . .” (Lynch, supra, 8 Cal.3d at p. 431, italics added.) The same is probably also true of the first. (See Pryor v. Municipal Court (1979) 25 Cal.3d 238, 244 [158 Cal.Rptr. 330, 599 P.2d 636] [saving disorderly conduct statute from vagueness challenge by construing “public place” to require actual or constructive knowledge “of the presence of persons who may be offended by the conduct”].) Traditionally the crime received relatively mild punishment, and apart from the potential magnifying effects of recidivism, that remains true today. (People v. Massicot (2002) 97 Cal.App.4th 920, 928 [118 Cal.Rptr.2d 705], citing Lynch, supra, 8 Cal.3d 410, 429-431.)
On the other hand, in the context of the assault case here the exposure charge may well have had the potential to inflame the jury against defendant in an unusual and peculiar manner. The entire defense consisted of raising doubts based on weaknesses in Gloria’s identification of defendant and his truck. As a repellent sexual aberration—what is still widely known as a “perversion”—the indecent exposure would naturally incline the jury to view defendant as a kind of freak, a pariah, a “pervert.” This would at a minimum reduce the jurors’ natural compunction about convicting him of the more serious offense on questionable evidence, as well as impairing their ability to view evidence of that offense objectively. Further, as discussed in more detail below, the prosecutor’s arguments to the jury tended to considerably amplify the inflammatory potential of the indecent exposure by furnishing the predicate to characterize defendant as a “predator” and “scary guy.” (See pts. IV.B., V., post.) We are satisfied that in the context of the heavily contested assault charges here, these arguments gave a particularly inflammatory tendency to the evidence of indecent exposure.
B. Weak Case Bolstered by a Strong Case
The indecent exposure case was extremely strong. The unimpeached testimony of the victim established that the exposure was perpetrated by a man who looked exactly like defendant and who was driving defendant’s car. The victim positively identified defendant from a photographic lineup the fairness and reliability of which was not questioned. There simply was no rational basis for the jury to doubt that it was in fact defendant who exposed himself to Gina. Defense counsel acknowledged the strength of this evidence and tacitly conceded her client’s guilt.17
*402In stark contrast to this seemingly unassailable showing, the evidence in the assault case was vulnerable to attack on a number of grounds. Indeed, while there was no reason to doubt that Gloria sincerely believed defendant was the assailant, virtually every part of her testimony on that subject presented discrepancies sufficient to lead a reasonable juror to entertain serious doubts about defendant’s guilt.
First, Gloria described her assailant as a Hispanic male. She denied using the term “Hispanic,” but acknowledged telling police that her assailant looked “Mexican American.” She testified that she used the term “Pocho,” meaning, “someone who was bom here and has Mexican parents.” She considered him “Mexican-American as opposed to just Mexican,” but the “American” component was not based on appearance; rather it rested on “the way he was speaking Spanish, that he didn’t speak it well.”
The record indicates that this attribution of a distinctly Hispanic appearance simply did not match defendant. The officer who filled out booking documents after defendant’s arrest recorded his ethnicity as “[w]hite.”18 The prosecutor elicited testimony from Detective Gutierrez that Gloria “identified” her assailant as “a light-skinned Hispanic male . . . .” But Gloria herself testified that the assailant’s skin was “light brown,” not “light,” and described it as “[sjimilar” to her own. Neither of the two mug shots in evidence shows defendant with a skin shade comparable to the distinctly olive tone in a photograph of Gloria. Instead they fully bear out the description of him as “white.” One shows him with an extremely pale complexion and sandy-*403colored hair. In the other neither his hair nor his complexion looks so pale, but his skin tone is far from what anyone would call “light brown.”
Nor was there any other evidence suggesting that defendant looked Hispanic. Indeed Gloria herself was not asked whether defendant, as he sat in the courtroom or was depicted in the photographs, looked “Pocho” to her. Further, while no testimony was sought from Gina Doe about the apparent ethnicity of the man who exposed himself to her, a telling inference may be based upon the photographic lineup from which she identified defendant. Of the six photographs selected by police for the array, undoubtedly based on Gina’s description, none had skin darker than one would expect in a person of northern European heritage. Only one had dark hair and eyes; and at least two—other than defendant—looked distinctly Nordic, Teutonic, or Slavic, with pale skin, blond hair, and light eyes.
In sum, so far as this record shows, Gloria’s assailant looked Hispanic, while defendant did not. This serious weakness in the prosecutor’s case is reflected in his repeated resort to speculation and surmise entirely outside the evidence. Thus he argued that a person’s ethnicity is often “a question of impression” and that some people defy ethnic categorization, implying that the assailant, and thus defendant, might have been “half Mexican and half white” or “a very light-skinned Hispanic . . . .” In doing so he ignored, and manifestly hoped the jury would ignore, Gloria’s own testimony that the assailant’s skin color was “light brown” and “similar” to hers, and that the only thing that seemed “American” about the assailant was his poor Spanish.
The prosecutor also insisted that there was “no evidence to say the defendant is not Hispanic,” and that apart from the booking officer’s entry, there was “no evidence presented in this case to say he is white . . . .” But apart from the issue of the burden of proof, the question was not defendant’s actual ethnicity, it was whether he looked like the assailant described by Gloria, and particularly as described by her before she decided defendant was the man. The prosecutor’s own insistence that a person can have Mexican genes and still look “white” was a tacit concession that defendant did look “white” and did not have, as Gloria said her assailant had, “light brown skin.” Nothing in this record supports a contrary view.
One point on which Gloria’s description of her assailant did nearly match defendant was her estimate that he stood about five feet nine inches tall, which is one inch shorter than defendant. The prosecutor laid great emphasis on the nearness of this estimate. But even here there was considerable room to doubt Gloria’s testimony. At trial she said she had told police her assailant was “between 5'6 to 5'8, more or less,” which would make him two to four *404inches shorter than defendant. Moreover a jury viewing the evidence dispassionately might observe that height was the assailant’s one outward feature Gloria would seem to have been least able to accurately judge. According to her, she was already sitting part way in her car when the assailant suddenly appeared at the driver’s door and “threw” her “back into the car,” apparently meaning further into the car. Thereafter neither of them was standing from the beginning of the assault until after she pushed herself out the opposite door, got up off the ground, and ran. The only indication that she ever saw the assailant while both were standing was her testimony that after seeking help at the bakery she “ran over to the comer and saw [the assailant] get into this track that he had parked there.” But she also testified that “the guy from the bakery was with me” at this time, and that person, Armondo Romero, described no such thing. All he saw was “a pickup truck pass by.”
The prosecution faced a further difficulty in Gloria’s description of the assailant to police as “[t]hin, skinny.” Defendant was certainly lean, apparently weighing about 180 pounds at the time of his arrest, but he was very far from “skinny.” Both photographs suggest the build of a wrestler. Indeed this is one of the more striking divergences between photographs of defendant and the police sketch, which depicts a distinctly thin-necked and narrow-faced person, in contrast to defendant’s bull neck and broad face. Other evidence—including the pictorial exhibits—suggested that defendant had some prominent physical features that Gloria never mentioned to officers, and did not at trial recall seeing on the assailant. These included distinctly protruding ears, a deeply furrowed brow, a receding hairline, and, apparently, “highlights” in his hair.
Another weakness in the prosecution’s case was the unexplained ineffectuality, and ultimate failure, of the assault. Although the record does not disclose Gloria’s height and weight, the prosecutor conceded in argument that she was shorter and less strong than defendant. At the same time, the jury heard testimony that defendant held a black belt in the martial art of Brazilian jujitsu, and was a high-ranked world-class contender in that sport as well the sport of “submission grappling.” Defendant’s hypothetical identity as Gloria’s assailant may thus seem at odds with her having out-grappled her assailant and fled the scene.
Gloria’s identification was also rendered vulnerable by her adamance coupled with a manifest eagerness to bring about the apprehension and punishment of her assailant. She told Detective Gutierrez that she was “100 percent sure that the defendant was the man who attacked her . . . .” But she acknowledged that after the attack she was afraid and “wanted the person *405who did this to [her] to be caught.” When she identified defendant in the in-field showup, she “had already decided at the car wash . . . that’s the person that attacked [her] . . . .” She had formed that conclusion while watching defendant, as she said, “from afar”—too far to see his eyes, height, weight, or whether he was Hispanic, or to read the truck’s license plate. She acknowledged that when she saw the black truck outside the carwash, she “got a little bit scared because it was a black truck.” But she also assented to assertions that she was “sure that that’s the truck and the person that attacked you,” “even though it was not a Ford Bronco,” and even though she could not see the driver’s height or weight, and that she “just felt sure that that was the person . . . .”
These features made Gloria’s testimony particularly vulnerable in light of the expert testimony of Professor Geoffrey Loftus, offered by the defense concerning the potential unreliability of eyewitness identifications. Compressed to its essentials, this testimony informed the jurors that reported memories may not only be rendered inaccurate by deficiencies in the witness’s original perception, but may in effect be corrupted by information later encountered and incorporated into memory. Thus, “every time you think about some event that you’ve experienced in the past, every time you revisit it in your memory, your memory will change in some fashion or another. . . . fit] • • • [f] • • • [A]s a result of all this kind of reconstructive processes, memory . . . can change . . . and does change over time, sometimes fairly dramatically. So that by the time a person is called upon to recount what happened to them, their recollections, their memory of what happened may be very different from the memory that they laid down at the time that the original event was taking place.”
Of particular relevance was the professor’s warning against reliance on a witness’s subjective confidence as an indicator of reliability. “[Tjhousands of experiments,” he said, show that “people can report memories . . . with a great deal of confidence . . . that are just dead wrong in many important respects.” “[Mjemory reconstruction” played a role in various criminal convictions based on mistaken identification, a “large majority” of which involved “faulty eyewitness testimony, witnesses identifying the defendant.. . often quite confidently, but incorrectly, as it turns out.” In some cases where people were exonerated on the basis of DNA evidence, they had been “misidentified by numerous eyewitnesses.” He opined that a witness’s adamance about the correctness of his or her identification should be completely disregarded in assessing the reliability of identification testimony.
Despite its generality—which was partly the result of a successful motion in limine by the prosecution—this testimony brought out several factors, not otherwise obvious to the jury, that could well cast Gloria’s identification into *406doubt. For example, it could highlight the serious deficiencies in the conditions under which she observed her assailant. Although she was certainly viewing him at close range throughout the struggle, the lighting consisted entirely of whatever illumination entered her car from the parking lot lights above it. Aside from its relative dimness, the overhead orientation of such lighting would naturally cast much or all of the assailant’s face into deep shadow. Moreover, Gloria testified that for much or all of the struggle, the assailant was pulling her head back by the hair. This would seem likely to direct her gaze upward and to make it more difficult to look directly at his face—unless his face itself was above her, in which case it would have been entirely in shadow. The possibility of “weapon focus” was also present, for she not only saw a knife but was told by the assailant that he had a gun.
Further serious weaknesses attended Gloria’s identification of defendant’s truck as matching that of the assailant. Gloria initially specified a make and model of truck—a “1986 Ford Bronco”—that simply did not match defendant’s 1981 Chevrolet Blazer. Although Gloria testified that she did not “know that much about the makes of cars,” she acknowledged having a friend who drove a Bronco. She herself drove a sport utility vehicle—a 2003 GMC Envoy—which she described as “my truck.”
Further, whatever Gloria knew or did not know about the names of various models, when she saw another Ford Bronco a week after the assault she took down its license number and told police she thought she had found the assailant’s vehicle.19 Thus, in order to find beyond a reasonable doubt that the assailant was driving defendant’s truck, the jury had to accept that Gloria was mistaken not only in initially describing a different vehicle but in subsequently recognizing a vehicle that matched her original description, but not defendant’s truck.
The prosecution sought to get over this difficulty by emphasizing the similar body styles of Blazers and Broncos. But apart from the debatable sufficiency of this consideration to meet the issue, its underlying premise, and Gloria’s truck identification in its entirety, were cast in doubt from an entirely independent source: the testimony of bakery worker Armondo Romero, the only other person to see the assailant’s vehicle, who described it not as a sport utility vehicle, but as a “pickup truck.” In her final comments to the jury, defense counsel cited this discrepancy as one of five respects in which the evidence raised a reasonable doubt as to defendant’s guilt. The prosecutor’s only response was to refer to defendant’s truck as a “pickup truck” and *407then to suggest in argument that both a Bronco and a Blazer are “basically, . . . pickup trucks with shells on the back.” It is unclear whether this was intended to confuse the jury, or perhaps to invite speculation that Mr. Romero had mistaken defendant’s Blazer for a pickup truck with a shell. When Mr. Romero was on the stand, the prosecutor had made no attempt to establish that the “pickup” he saw might have had a shell. By alluding to such a possibility in jury argument, he was simply fabricating facts out of thin air, and inviting the jurors to do likewise.
These circumstances—most notably Gloria’s contested original description of the vehicle, its later metamorphosis from a Bronco to a Blazer, the questionable viewing conditions preceding her “decision” that defendant was the assailant, and the highly suggestive circumstances of her formal identifications—would have suggested to a reasonable juror, versed in the testimony of Professor Loftus, that her identification of defendant and his truck could well rest on after-the-fact reconstructions rather than accurately retrieved original perceptions. As defense counsel pointed out, everything Gloria remembered about the assailant and his vehicle seemed to be revised in a flash two and a half weeks after the event, when she saw defendant sitting in his Blazer and formed the conviction that he was the culprit. At that moment, a skeptical juror could quite reasonably conclude, the assailant metamorphosed from Hispanic to European, from light brown to white, from skinny to muscular; his vehicle metamorphosed from a 1986 Ford to a 1981 Chevrolet; and its rear window vanished. This is not the stuff of strong cases.20
V. Countervailing Benefits
Once the spillover evidence has been identified and its potential to influence the verdict evaluated, the trial court must weigh the resulting risk of prejudice against the advantages of joint trials. (Soper, supra, 45 Cal.4th at p. 775.) The greatest advantages will often be those stemming from the avoidance of duplication of evidence, as where at least some of the same testimony, or testimony from the same witnesses, would be heard in each of *408the hypothetical separate trials. (See Alcala v. Superior Court, supra, 43 Cal.4th 1205, 1218 [“ 1 “joinder prevents repetition of evidence and saves time and expense to the state as well as to the defendant” ’ ”].) In these cases, joint trials may serve not only the interests of the state, but those of third persons, most obviously by sparing one or more citizen witnesses from having to testify more than once.
Even where no case-specific economies appear, we are directed by the Supreme Court to weigh, as against the risk of prejudice, the “systemic economies” of joint trials, i.e., the savings in costs otherwise incurred by the state in a second trial and appeal. (Soper, supra, 45 Cal.4th at p. 782; see id. at pp. 772, 781-783.) We do not understand the court to mean that these “systemic economies” will always be enough, by themselves, to sustain a refusal to sever. Rather the court was explaining its conclusion that the Court of Appeal there had “inappropriately minimized the benefits of joinder” when it dismissed the benefits as “minimal.” (Soper, supra, 45 Cal.4th at p. 781.) The court’s point was that reviewing courts must not take lightly the public benefits that flow from the administrative savings that flow from disposing of multiple charges in a single proceeding. (See Soper, supra, 45 Cal.4th at p. 782.)
Here the “systemic economies” were the only advantages of joint trial. There was absolutely no overlap in the evidence or witnesses. So far as the evidence was concerned, the matter essentially took the form of two trials before a single jury. The time spent actually trying the indecent exposure was minimal; the relevant testimony takes up approximately 22 pages of transcript. Therefore the only apparent savings was the time and cost to impanel a second jury to hear that evidence separately. This is not a trivial matter; in all likelihood, it would have consumed more time than the actual taking of evidence, and probably the deliberations as well. Moreover we do not disregard the fact that however honorable and necessary jury duty may be, it incommodes 12 citizens who would otherwise remain free to pursue their own concerns.21
Ponderable as these interests are, however, we do not believe they can counterbalance the very substantial risk that evidence of the indecent exposure played a dispositive role in the verdict on the assault charge. Obviously a balance must be struck between economy and convenience on the one hand, and accuracy and fairness on the other. Each of the signatories of this opinion has served as the presiding judge of a major metropolitan trial court. In that role we have been responsible for the efficient allocation of resources that never seemed to keep pace with our court’s burgeoning *409caseload. We are therefore far from insensitive to concerns about the economical use of those resources. But it would never occur to us, and we trust it would never occur to any judge, to elevate those concerns above an accused person’s right to a fair trial—including the right to be tried (and convicted, if that is the result) only upon evidence the law allows the fact finder to consider. Such an ordering of priorities would be irreconcilable with the basic consensus on which our society rests, and would strike a blow at the requirement of due process of law which is the beating heart of our republic. We are confident that the Supreme Court had no such intention when it admonished reviewing courts to take due account of “systemic economies.” Those economies simply were not enough here to justify a joint trial.
There is no suggestion that, with the possible exception of the actual jury arguments made by the prosecutor, any of the factors we have identified were unforeseeable at the time the trial court made its ruling. The basic outlines of the case, and the general use the prosecution made of the indecent exposure, were apparent from the preliminary hearing onwards, and were accurately forecast by defense counsel in support of the motion for separate trials. Because we do not believe that a reasoned discretion could reach any other conclusion than that separate trials were necessary to avoid undue prejudice, we hold that the trial court abused its discretion by denying the motion seeking such trials.
VI. Due Process; Actual Prejudice
Even where the information of record fails to show that the trial court abused its discretion in denying a motion for separate trials, we are still obliged to reverse if the appellate record shows that “ ‘the joinder of counts or defendants for trial resulted in gross unfairness depriving the defendant of due process of law.’ ” (Soper, supra, 45 Cal.4th at p. 783, quoting People v. Rogers (2006) 39 Cal.4th 826, 851 [48 Cal.Rptr.3d 1, 141 P.3d 135].) Although this rule places a “high burden” on the appealing defendant (Soper, supra, 45 Cal.4th at p. 783), the present record satisfies that burden. Having reviewed it with some care, we find compelling evidence that the joint trials inflicted actual prejudice upon defendant. Indeed, we find a considerable likelihood that had the jury not heard evidence of the indecent exposure, it would not have found defendant guilty of the assault.
Short of statements by the jurors themselves, nothing could more vividly demonstrate the crucial role played by the indecent exposure in securing the assault conviction than the prosecutor’s own heavy and pervasive reliance on it in jury argument. It is no exaggeration to say that he mentioned the indecent exposure at every opportunity, on every conceivable pretext, and for every possible purpose. In his opening argument he characterized the exposure incident as “[p]robably the most powerful evidence” that defendant was *410Gloria’s assailant. He urged the jury to view it as the equivalent of genetic evidence, asserting, “We don’t have any DNA . . . [b]ut what we do have is powerful powerful corroboration. And that is the defendant’s indecent exposure three months prior to the assault.” (Italics added.) He repeated the point a few minutes later, saying, “We don’t have DNA but we do have the fact he committed this 314 [i.e., indecent exposure].”
Next he told the jurors that criminal exhibitionism furnishes evidence of propensity to commit rape. This argument rested not on evidence, but on an invented psychiatric reality. Thus he invited jurors to “think, what’s the state of mind of a guy who’s going to be mast[u]rbating in front of a woman and get her attention drawn to himself[?]” Before the jurors could actually follow this suggestion to speculate on their own, or perhaps to realize that they had no idea what goes on in the mind of a criminal exhibitionist, he helpfully supplied his own inflammatory surmise: “So we go into this whole analysis as to whether or not he’s the perpetrator of the assault to commit rape from the premise that this is a man who is mast[u]rbating in front of a woman, who called attention to her—to himself, three months before, [f] This man is a predator. He’s building up. He’s starting small. He exposes himself. [][] Is it really a stretch to think that his next step, three months later, because he gets so agitated, for whatever reason, whatever his state of mind, is he decides to commit a sexual assault[?]” (Italics added.)
We have already adverted to the inherent potential of the indecent exposure to convey an inflammatory impression of defendant as a deviant or pervert. In addition, the power of the term “predator” to incite passion and prejudice can hardly be overstated. The term connotes, if it does not denote, one who habitually commits sex offenses characterized by violence, pedophilia, or both. Thus, with one inconsequential exception, California prescriptive law speaks only of sexually violent predators, a phrase defined to include pedophiles. (See Welf. & Inst. Code, § 6600 et seq. [Sexually Violent Predator Act (SVPA)]; cf. Ballot Pamp., Gen. Elec. (Nov. 7, 2006) text of Prop. 83, p. 127 [initiative amending SVPA entitled “The Sexual Predator Punishment and Control Act: Jessica’s Law”].) The distinction between violent and nonviolent sex crimes is recognized throughout the relevant literature. One commentary defines “ ‘sexual aggressors’ ” to include, as relevant here, “all those who have engaged in sexual acts involving physical contact with a child or nonconsensual sexually violent acts against adults (i.e., offenses involving the use of force in such a way that the sexual integrity of the victim was violated).” (2 Faigman et al., Modem Scientific Evidence: The Law and Science of Expert Testimony (2008) § 11.24, pp. 203-204, italics added.)
Defense counsel quite properly objected to the prosecutor’s psychiatric speculations, stating, “There’s no evidence as to this argument.” The court *411not only overruled the objection, but invited the jury to give the prosecutor’s argument whatever credence it chose: “The jury has already been admonished that evidence is to be decided from the testimony of witnesses. You have some latitude with respect to argument, [f] They can give it whatever weight they want.” (Italics added.)
Shortly thereafter the prosecutor returned to the theme that, based upon the indecent exposure, defendant was a loathsome and dangerous pariah. On the pretext of discussing motive, he asserted that defendant “is a sexual deviant because he mast[u]rbated in front of [Gina Doe].” (Italics added.) He went on to characterize defendant, based solely on the indecent exposure, as “a scary guy” and, again, a “predator.”
These arguments—and thus the indecent exposure evidence—struck at the heart of the identity defense by inviting the jury to substitute anti-“deviant” bias for a reasoned analysis of the evidence. Virtually every time the prosecutor acknowledged the question of identity he cited or alluded to the indecent exposure as an answer. Just as predicted in the defense motion for separate trials, he repeatedly lumped the two offenses together in the manifest hope that the certainty of defendant’s commission of the lesser crime would cure the uncertainties attending the greater. Thus he argued, “So how do we know it was the defendant^] How do we know that the man in court today is the man who committed these crimes against these two women? [f] Well, [Gina Doe’s] identification of the defendant [for the indecent exposure] has gone completely unimpeached .... There’s no dispute about that. [][]... I’d be surprised if the defense made an argument that he was not the one that committed the . . . indecent exposure on [Gina].” Shortly thereafter, he explicitly invited the jury to consider the indecent exposure as evidence that defendant was the assailant: “[I]n making your decision, you need to consider all of the evidence. And you can consider all of the evidence that was presented in this trial, every piece of evidence. You can consider the facts behind what happened to [Gina Doe] in deciding the offenses in .. . the assault cases.” (Italics added.)
Because there was ample reason for a jury to entertain a reasonable doubt that Gloria had correctly identified her assailant, it is entirely possible, if not likely, that a jury not exposed to evidence of the indecent exposure incident, and the use the prosecutor made of it, would have acquitted defendant of the assault. We must therefore conclude that the trial court’s denial of the motion for separate trials, if not an abuse of discretion, resulted in such gross unfairness as to deprive defendant of a fair trial and thus of due process of law. This conclusion makes it unnecessary to address defendant’s other claims of error or to address his petition for habeas corpus.
*412Disposition
In No. H031525, the judgment of conviction on the charge of indecent exposure is affirmed. The judgment on the counts charging assault is reversed for further proceedings in accordance with this opinion. In No. H032982, the petition is dismissed as moot.
Premo, J., concurred.
Both victims happen to have the same initials. This makes it impossible to discuss the facts intelligibly using initials only. We will therefore refer to the victims by fictitious names. Insofar as an order is required for this purpose, such an order is hereby made.
She testified that he held a small knife and that after she escaped she discovered a cut on her hand. A photograph in evidence depicted a cut on her outer wrist. This evidence formed the basis for the charge of assault with a deadly weapon. Police found a “small retractable knife” in the living room of defendant’s home, which he shared with his mother and brother. That knife tested negative for blood. It was not offered in evidence and was never shown to Gloria.
The prosecutor told jurors that the assailant also used “a small retractable knife,” but no evidence supported the second adjective. Gloria only said the assailant had “a little small, like a hand, a hand knife,” which was “very narrow.” She was unable to say whether it was a “switchblade.” Implicitly acknowledging the transparency of his mischaracterization, the *381prosecutor allowed that “maybe” defendant “used a different knife” than the one found by police, but urged the jurors to still draw an incriminating inference from finding of a knife in his shared home.
After defendant’s arrest, Detective Gutierrez retrieved a black garment from defendant’s home through defendant’s girlfriend. On questioning by the prosecutor the detective characterized this garment as a “black jacket....’’ On further defense examination he allowed that it was in fact a “wool blazer . . . .”
For the ostensible purpose of illustrating the similarity of the two vehicles, the prosecutor introduced three small photographs of defendant’s truck and a low-resolution printout of an advertisement for a 1981 Ford Bronco. In argument he told the jury that the vehicles thus depicted bore a striking resemblance to one another: “The lights are similar. Really the only difference, there’s a minor difference in the grill and that’s a Chevy logo versus a Ford logo.” The record does not disclose whether a similar degree of resemblance might have been found between defendant’s truck and the model actually described by the witness.
Gloria testified that upon her truck’s emergence from the carwash, the man in the truck “t[ook] off all of the sudden, really fast and dr[o]ve up next to where my truck had come out of the car wash and pull[ed] up right next to it, lookfed] at it and then continue[d] driving.” A juror viewing Gloria’s testimony critically might view this account with skepticism, since nothing in the record can explain the conduct thus described. At the time of the assault Gloria was driving a four-door sedan, not the Envoy that supposedly triggered this bizarre reaction from the Blazer driver. Her assailant had no apparent reason to take any notice of the Envoy.
The indecent exposure was originally charged in a separate file and later consolidated with the assault charges. The file of the indecent exposure case prior to consolidation does not appear in the present record.
The abstract of judgment omits the charge of assault with a deadly weapon, but the omission appears immaterial since the court’s minutes reflect an intention to stay sentence on that count pursuant to Penal Code section 654. The abstract of judgment also appears mistaken in stating that the court stayed sentence on the indecent exposure count under that statute. Such a stay would be plain error since the charges arise from entirely distinct events. The court’s intention was to impose a six-month term on the indecent exposure to be served concurrently with the assault.
Because counts 1 and 2 arose from a single instance of conduct, we will often refer to them in the singular as “the assault charge.”
One study showed that 19 percent of exhibitionists reported committing rape, while 15.4 percent of rapists reported engaging in exhibitionism. Another study yielded corresponding figures of 28 percent and 25 percent.
The prosecutor declined to lodge an objection to the Abbott declaration, expressing confidence at the hearing that the court would “give it whatever weight the court feels is appropriate.”
Our dissenting colleague’s analysis of eligibility exemplifies the ongoing judicial struggle to give the statute a coherent meaning and application. The statute’s primary criteria—that crimes be “connected together in their commission” or belong to the “same class of . . . offenses” (§ 954)—are highly ambiguous. Their unworkability as guides for decision has *387caused them to be supplanted by judicial glosses that are themselves remarkably vague. (See People v. Kemp (1961) 55 Cal.2d 458, 476 [11 Cal.Rptr. 361, 359 P.2d 913] [“ ' “ ‘same class of crimes’ ” ’ ” means “ ‘ “offenses possessing common characteristics or attributes” ’ ”]; People v. Matson (1974) 13 Cal.3d 35, 39 [117 Cal.Rptr. 664, 528 P.2d 752] [“ ‘connected together in their commission’ ” requires a “ ‘common element of substantial importance’ ”].) The dissent insists that these nebulous formulae describe two distinct tests, but the only apparent difference is that the first is more overbroad than the second, logically subsuming it. Among the statute’s other weaknesses is its failure to distinguish between the joinder of multiple charges against a single defendant, as to which the state should obviously enjoy greater latitude, and joinder of charges against multiple defendants, as to which a more restrictive rule seems necessary and appropriate. (Cf. Fed. Rules Crim.Proc., rule 8, 18 U.S.C.)
Counsel wrote in part, “In the present case, failure to sever the counts against Mr. Earl[e] would be unduly prejudicial. The evidence of the two offenses would not be cross-admissible. The two incidents occurred three months apart and were in no way connected to one another. The case involving indecent exposure occurred during the day and there was no inference of force, violence or a weapon. The second offense alleges a forceful and violent attack with a knife at 5:00 in the morning. There is no physical evidence linking Mr. Earl[e] to this incident. The incidents occurred] at different locations, different times of the day and involve very different conduct.”
The dissent denies that there was any evidence that the assailant “attempted to touch, [Gloria’s] private parts” or “to remove her clothing.” (Dis. opn., post, at p. 420.) This may be narrowly true: When Gloria was asked whether the assailant “tr[ied] to take your pants off,” her reply was, “No.” But she went on immediately to explain, “I don’t think he was—he had a chance to do any touching, because he was still trying to make me lie down and he was pulling my hair. So that’s what he was doing with his hands.” She then testified that she could not remember whether he tried to unbutton her pants. Shortly before this she had testified, “I noticed that all of the sudden he seemed to become even more aggressive. I had fairly tight pants on and I also had an apron on over them that was tied[,] [a]nd this seemed to infuriate him even more.” The unmistakable purport of this testimony was that if defendant did not succeed in removing her clothing and touching her intimately, it was not for want of trying, but only because she fought off his attempts to subdue her to a point where he could accomplish his obvious goal. No rational juror could have supposed otherwise.
Although the dissent does not defend the use of the evidence to show identity, it does attempt to convey the impression that the two offenses had small details in common. A good example is its statement that defendant “summoned [Gina] to his car window” whereas the assailant “approached the car window of another lone female stranger;” i.e., Gloria. (Dis. opn., post, at p. 420, italics added.) The first statement is an accurate, characterization of the testimony. The second is not, though its literal truth may be deduced from Gloria’s testimony that the assailant appeared while she was sitting in her open car door. By approaching the car, the assailant perforce approached all of its component parts, including its windows. One might say with equal accuracy—and equal pertinence—that he approached the steering wheel or the battery.
Evidence Code section 1108 would not authorize the admission of expert opinion concerning the presence or absence of a particular disposition in a particular defendant. (People v. McFarland (2000) 78 Cal.App.4th 489, 495 [92 Cal.Rptr.2d 884]; see id. at p. 496, citing Pen. Code, § 29 [expert may not testify on “the ultimate question of whether the defendant had or did not have a particular mental state at the time he committed the offense”].) This does not mean that in a sex crimes prosecution, the state may not be required to lay a proper foundation for an inference of propensity to commit the charged offense. For instance, testimony like Dr. Abbott’s, addressing the incidence of rapes by criminal exhibitionists and vice versa, and the factors bearing on the likelihood of an exhibitionist’s committing rape, would offend neither of the prohibitions just noted, but would provide the jury with an evidentiary foundation on which to predicate an inference, as well as an estimate of the likelihood, that defendant’s commission of indecent exposure actually reflected a propensity to commit rape. It was not defendant’s burden to make this showing. It is for the proponent of evidence to establish the foundational facts for its admission, including its relevance to a material issue. (See Evid. Code, § 403, subd. (a).)
To be sure, nonsexual crimes may also be the product of psychological compulsion. “Kleptophilia” describes a tendency to receive sexual arousal or gratification from theft or burglary. (Dorland’s Illustrated Medical Diet. (29th ed. 2000) p. 947.) One who has exhibited this trait might be expected to commit crimes of larceny and burglary in the same way that a criminal exhibitionist might be expected to repeat his offense. Indeed the compulsion need not *399have sexual origins; it may be driven, as in the case of “kleptomania,” by a compulsion or impulse disorder. (See American Psychiatric Assn., Diagnostic and Statistical Manual of Mental Disorders (4th ed. text rev. 2000) p. 667.) Of course, Evidence Code section 1108 would not apply, and this evidence would presumably not be admissible, in a prosecution for theft or burglary.
The dissent states flatly that defense counsel “conceded in closing argument that defendant was guilty of the indecent exposure count,” which conveys the impression that counsel expressly acceded to defendant’s guilt. (Dis. opn., post, at p. 416.) She did not. Her *402relevant comments were as follows: “Now, I want to show briefly about what I believe the evidence has shown in this case as to count three [the indecent exposure]. The evidence is there, I expect you to go back and do your duty and vote appropriately. [][] According to the evidence with regard to counts one and two, I’m going to submit to you, based on the evidence, Mr. Earle is not guilty. And I’m going to show you why I believe that. []Q And I wanted to mention again about count three because we heard that count three corroborates counts one and two. And there is no evidence that count three corroborates anything, count three is what it is. And there is no evidence, no nothing to suggest that a person who sits in a car and exposes himself—unsavory though it may be—is a person who would attack someone in the middle of the night using a knife. [H And I say that because there was no evidence to suggest that. And just because it’s convenient to say he did count one and two because he may have done three, convenient, is not what this is about. This is about a process and that’s why you’re here. Q] There was some talk about, perhaps, after the exposure he was frustrated and he was looking—there is no evidence as to that. That’s all speculation and you’re not allowed to use that.” (Italics added.)
In jury argument the prosecutor sought to minimize this characterization by calling it “the impression of the person who took the booking photo.” But the prosecution itself relied on nothing more than the “impression” formed by an untrained, terrified observer fighting off a nocturnal assault inside a vehicle illuminated only by overhead parking lot lights. This lends considerable irony to the dismissal of the officially recorded “impression” of a (presumably trained) police officer, formed under conditions suitable for photography.
Curiously, when a police officer was asked how authorities concluded that the driver of the Bronco was not a suspect, he replied only, “I believe we didn’t find any photos on this individual so we couldn’t conduct any type of lineup.” This suggests not that police ruled him out, but that they failed to investigate further, perhaps because of defendant’s arrest.
The dissent makes no attempt to identify any inaccuracy in our account of the prosecution evidence but responds that the jurors rejected defense attacks on that evidence. Of course they did; otherwise they would have acquitted defendant, and the case would not be before us. The question here is whether the record reveals an unacceptable likelihood that they were led to their decision by evidence of the indecent exposure. The only point of any substance made by the defense is that defense counsel conceded a similarity between the police sketch and an older mug shot of defendant—not the mug shot taken after his arrest for the assault, less than three weeks after Gloria supposedly saw him. It is true that the sketch bears some resemblance to defendant as depicted in that mug shot, but it does not impress us as a striking one, and certainly need not have been found striking by a reasonable juror whose reasoning processes had not been poisoned by irrelevant evidence of other sexual misconduct and specious but inflammatory prosecution arguments predicated on that evidence.
We make nothing of the fact that in a misdemeanor case, as this would have been, the parties are free to agree to a jury of less than 12 persons. (Cal. Const., art. I, § 16.)