TITONE, J.

In 1984, the Legislature repealed the statutory provisions requiring corroboration of the victim's testimony in certain sex-crime prosecutions involving underage victims (L 1984, ch 89, amending Penal Law former §§ 130.16, 260.11). We now hold that the Ex Post Facto Clause of the United States Constitution (US Const, art I, § 10; see also, art I, § 9, cl 3) does not preclude the application of this repealing enactment to prosecutions for crimes occurring before its effective date. Accordingly, in this prosecution for various sex-related crimes committed against young children, there was no constitutional impediment to applying the new statute and convicting defendant solely on the basis of the victims' testimony, even though the charged crimes occurred before the repealing legislation became effective. However, defendant's convictions must nonetheless be reversed, because two prejudicial trial errors tainted the fairness of his trial.

FACTUAL BACKGROUND

At the time of the alleged criminal incidents, defendant was a remedial math teacher at Glenwood Elementary School in Vestal, New York. Official concern about his classroom conduct was aroused when the younger brother of one of defendant's former pupils, Matthew L., told his mother that he had heard about a teacher who puts his hands down boys' pants. When questioned by his mother, Matthew confirmed that he had been fondled by defendant and that defendant had put his hands down other boys' pants as well.

Understandably upset by this revelation, Mrs. L. sought to confirm the allegation by contacting the mother of another boy, Jason A., who had been in defendant's class with Matthew. Matthew and Jason apparently discussed the matter on the school bus the next morning, for rumors about defendant's misconduct had already spread to the school and reached the principal's ears by the time Mrs. L. contacted him. Within a few weeks, the police became involved and launched a wider investigation into defendant's conduct. Within a three-week period, the police had obtained inculpatory evidence from eight boys, all of whom were defendant's students.

As a result, defendant was indicted on 23 counts of first degree sexual abuse and 9 counts of endangering the welfare of a child, all involving children under the age of 11. Seven of the sexual abuse counts were dismissed for either duplicity or lack of specificity. Of the remaining 25 counts, 19 involved incidents occurring between September 1983 and November 1984 and 6 involved incidents occurring between November 1984 and December 1985.

Before the commencement of trial, the trial court denied a CPL 190.65 motion by defendant to dismiss the 19 counts involving pre-November 1984 incidents on the ground that those counts were not supported by legally sufficient Grand Jury evidence. The motion was based on a defense contention that the Ex Post Facto Clause of the United States Constitution (US Const, art I, § 10, cl 1; see also, art I, § 9, cl 3) foreclosed application of the recently enacted statute repealing the corroboration rules that would otherwise have governed the trial on those counts (see, L 1984, ch 89, amending Penal Law former §§ 130.16, 260.10).

At defendant's trial, which commenced on June 17, 1986, all eight of the children who had been interviewed by the police, as well as a ninth child, testified. With one exception, each boy described, in essentially similar terms, how he had stood behind defendant's desk and had his penis touched or fondled by defendant while ostensibly receiving help with a math problem. David A., who was not himself abused, stated that, upon approaching defendant's desk, he had seen defendant with his hand down Matthew L.'s pants.

Defense counsel closely cross-examined each of the prosecution witnesses as to whether he had discussed, or heard other students discuss, the allegations about defendant or had heard anything about the police investigation of defendant's conduct. Additionally, each witness was questioned about his failure to report the alleged incidents at the time they occurred, as well as whether he liked or disliked defendant and whether he thought or heard that defendant was gay. These lines of questioning were intended to support the defense's primary theory, first advanced in counsel's opening statement, that a false accusation had begun when Matthew lied to his younger brother about defendant's conduct and then felt a need to adhere to his story when questioned to avoid punishment. It was the defense's position that after Matthew's story first came to light it had snowballed, quickly spreading throughout the school and inciting other students to make similar — and equally unfounded — accusations.

In response to this defense contention, the People were permitted, over vigorous objection by the defense, to call an additional witness, one Domenick M., who had been in one of defendant's classes but had moved to another State more than a year before the investigation into the present charges began. This witness testified that, like the others, he had been fondled by defendant while he was supposedly receiving help with a math problem at defendant's desk. He further testified, in contrast to the others, that he had promptly reported the matter to his grandmother, who, in turn, had reported the matter to his mother. The grandmother also was permitted to testify about Domenick M.'s statements to her, her own statements to the boy's mother and the letter that the mother had sent to defendant and the school psychologist in which she requested that the improper touching be stopped.

Although the trial court initially had misgivings about admitting this evidence of a prior uncharged crime, it ultimately determined that the testimony was admissible by analogy to the rule governing the use of a witness's prior consistent statements. In the trial court's view, Domenick M.'s testimony was admissible because it tended to rebut the defense theory that the current charges were the product of either a rumor or a conspiracy among the children who were in the school at the time Matthew L.'s accusation first surfaced.

After the prosecution rested, the defense called 12 witnesses, including defendant, several character witnesses and several teachers, who testified that they had never witnessed any sexual misconduct on defendant's part although they often had occasion to be in his classroom. Defendant was not permitted, however, to call the two investigators, Officer Barnard and Investigator Kintner, who had originally interviewed the young victims. Defense counsel unsuccessfully argued that he should be permitted to question the officers about these initial interviews so that he could elicit that the boys' testimony had been tainted by police suggestion. In support of his position, counsel pointed to the evidence that at least one of the boys, Kevin F., had first denied that any improper touching had occurred and then changed his mind after Investigator Kintner told him that the police had two witnesses who had seen such touching occur. Counsel also represented, based upon the police reports he had seen, that "there may be three or four or five other such incidents." Nonetheless, the trial court concluded that the proposed questioning was relevant only to the victims' credibility and was therefore "collateral." As a consequence, the court excluded all questioning of the investigating officers, with the exception of a brief series of leading questions concerning Investigator Kintner's interview with Kevin F.

The jury found defendant guilty of all of the submitted charges. The Appellate Division, in a 3-2 decision, affirmed the judgment of conviction, holding that the present law did not require corroboration of the underage victims' testimony and that the Ex Post Facto Clause of the Constitution did not foreclose application of the present law to charges involving crimes that had been committed before the law's enactment. A majority of the court also rejected defendant's arguments concerning the admission of Domenick M.'s testimony and the restriction of his right to question Officer Barnard and Investigator Kintner. Defendant was granted leave to take a further appeal by one of the dissenters at the Appellate Division.

DISCUSSION

I. Ex Post Facto

At the time defendant allegedly committed the crimes for which he was tried, Penal Law § 130.16 provided:

"A person shall not be convicted of consensual sodomy, or an attempt to commit same, or of any offense defined in [article 130] of which lack of consent is an element but results solely from incapacity to consent because of the alleged victim's age, mental defect, or mental incapacity * * * solely on the testimony of the alleged victim, unsupported by other evidence tending to:

"(a) Establish that an attempt was made to engage the alleged victim in sexual intercourse, deviate sexual intercourse, or sexual contact, as the case may be, at the time of the alleged occurrence; and

"(b) Connect the defendant with the commission of the offense or attempted offense."

Penal Law former § 260.11 (amended L 1984, ch 89) applied the same corroboration requirement to the crime of endangering the welfare of a minor.

Effective November 1, 1984, the Legislature amended both statutes so that corroboration was no longer required in cases where the incapacity to consent resulted solely from the victim's age. In defendant's case, which involved 19 counts charging repeated sexual misconduct toward children occurring before November 1, 1984, most of the charges were supported only by the testimony of the underage victims. Thus, defendant could not properly have been convicted on those 19 counts under the former statutes, and we are squarely confronted with the question whether the amended versions of Penal Law §§ 130.16 and 260.11 may be applied to his case without running afoul of the constitutional prohibition against ex post facto laws.

The modern formulation of the ex post facto rule may be found in the three leading Supreme Court cases, Weaver v Graham ( 450 U.S. 24), Dobbert v Florida ( 432 U.S. 282) and Beazell v Ohio ( 269 U.S. 167). One purpose of the rule is to assure that citizens have fair warning of what conduct will be punished so that they may rely on the explicit laws and tailor their conduct accordingly (Weaver v Graham, supra, at 28-29; Dobbert v Florida, supra, at 297). The ban on ex post facto laws also serves to restrain "arbitrary and potentially vindictive legislation" (Weaver v Graham, supra, at 29; accord, Malloy v South Carolina, 237 U.S. 180, 183; Kring v Missouri, 107 U.S. 221, 229).

The critical elements necessary to establishing that a criminal or penal law is ex post facto are its retrospectivity and its detrimental effect on the accused (Weaver v Graham, supra, at 29; Dobbert v Florida, supra, at 297; Lindsey v Washington, 301 U.S. 397, 401). However, not every retrospective application of a new enactment that impairs a criminal defendant's rights is violative of the Ex Post Facto Clause. The constitutional prohibition extends to "any statute which punishes as a crime an act previously committed, which was innocent when done; which makes more burdensome the punishment for a crime, after its commission, or which deprives one charged with crime of any defense available according to law at the time when the act was committed" (Beazell v Ohio, supra, at 169). Further, "the constitutional provision was intended to secure substantial personal rights against arbitrary and oppressive legislation * * * and not to limit the legislative control of remedies and modes of procedure which do not affect matters of substance" (Beazell v Ohio, supra, at 171). Thus, "[e]ven though it may work to the disadvantage of a defendant, a procedural change is not ex post facto" (Dobbert v Florida, supra, at 293 [emphasis supplied]).

If this distinction between procedural and substantive changes in the law were mechanically applied, there would be no doubt that the repeal of corroboration laws would not be encompassed within the Ex Post Facto Clause, since such laws do not add to the legislatively defined crimes, do not alter the elements of existing crimes, do not elevate the punishment for existing crimes and are generally treated as "procedural" rules of evidence (see, People v Linzy, 31 N.Y.2d 99, 102-103; People v Blank, 283 N.Y. 526, 528). However, the seemingly simple distinction between matters of mere procedure and matters of substantive penal significance has not been so easy to apply in practice. As was recognized in Thompson v Utah ( 170 U.S. 343, 354-355), labels are not particularly helpful in this area of constitutional analysis, since changes affecting the penalty for or the substance of a crime may easily be disguised as procedural enactments (see, e.g., Miller v Florida, 482 U.S. 423; Weaver v Graham, supra; see also, Gibson v Mississippi, 162 U.S. 565, 590). Further, some procedural changes may have such a profound detrimental impact on the defense that they cannot fairly be disregarded (see, e.g., Kring v Missouri, supra). Consequently, the courts have struggled to define the types of changes in the law that, although labeled "procedural," will nonetheless be treated as ex post facto laws if given retrospective effect (see, Beazell v Ohio, supra, at 171).

In this regard, there have been two discernible lines of analysis. The first of these considers so-called "procedural" changes to be ex post facto if they impair a "substantial right" of the defendant (see, Weaver v Graham, supra, at 29, n 12; see generally, Tribe, American Constitutional Law § 10-3). The Supreme Court has used this approach to invalidate retroactive application of new procedural rules on only two occasions (Kring v Missouri, supra [invalidating retroactive use of former rule absolutely prohibiting retrial on higher count after conviction of lesser included offense] [described as a "substantial right" case in Hopt v Utah, 110 U.S. 574, 589; and Thompson v Missouri, 171 U.S. 380, 384-385]; Thompson v Utah, supra [invalidating reduction in number of jurors necessary for felony trial, as applied to crime committed before change in law occurred]). However, the court has twice rejected ex post facto challenges based on the "substantial right" theory to changes in evidentiary rules governing the admissibility of certain classes of evidence (see, Hopt v Utah, 110 U.S. 574, 590, supra [felons' testimony]; Thompson v Missouri, supra, at 386-388 [handwriting authentication]; see generally, Note, Ex Post Facto Limitations on Changes in Evidentiary Law: Repeal of Accomplice Corroboration Requirements, 55 Fordham L Rev 1191). A fortiori, the theory, whose scope is generally somewhat elusive, has no application in this case, which concerns a change in a rule requiring corroborative proof for what is otherwise competent and admissible testimonial evidence.

The second available approach has its origins in Justice Chase's seminal opinion in Calder v Bull (3 Dallas [3 US] 386, 390 [emphasis supplied]), in which he stated that the concept of an ex post facto law includes "[e]very law that alters the legal rules of evidence, and receives less, or different, testimony, than the law required at the time of the commission of the offence, in order to convict the offender." In Hopt v Utah (supra, at 590 [emphasis supplied]), the Supreme Court reiterated, in dictum, this concern about retroactive reductions in the quantum of evidence required: "Any statutory alteration of the legal rules of evidence which would authorize conviction upon less proof, in amount or degree, than was required when the offence was committed, might, in respect of that offence, be obnoxious to the constitutional inhibition upon ex post facto laws. But alterations which do not increase the punishment, nor change the ingredients of the offence or the ultimate facts necessary to establish guilt, but — leaving untouched the nature of the crime and the amount or degree of proof essential to conviction * * * relate to modes of procedure only * * *. Such regulations of the mode in which the facts constituting guilt may be placed before the jury, can be made applicable to prosecutions or trials thereafter had, without reference to the date of the commission of the offence charged."

While the Supreme Court has often cited this dictum, and the discussion in Calder (supra) on which it was based (see, e.g., Dobbert v Florida, supra, at 294; Thompson v Missouri, supra, at 385-386), the court has never actually applied it to invalidate a retrospective change in an evidentiary rule (see, Note, Ex Post Facto Limitations on Changes in Evidentiary Law, op. cit.). Nonetheless, the Hopt-Calder dictum has been relied upon by other courts as a basis for invalidating retrospective application of modern statutes abolishing outdated corroboration rules (Government of Virgin Is. v Civil, 591 F.2d 255; Bowyer v United States, 422 A.2d 973 [DC Ct App]; State v Schreuder, 726 P.2d 1215 [Utah]; State v Byers, 102 Idaho 159, 627 P.2d 788; but see, Murphy v Sowders, 801 F.2d 205 ; Commonwealth v Phillips, 655 S.W.2d 6 [Ky]; Murphy v Commonwealth, 652 S.W.2d 69 [Ky]; see also, Murphy v Kentucky, 465 U.S. 1072 [White, J., dissenting from denial of certiorari in Commonwealth v Phillips, supra; and Murphy v Commonwealth, supra]). We conclude, however, that this reliance is misplaced and that, to the extent that the Calder-Hopt reference to the quantum of proof is still viable, it has no application to the repealing legislation at issue here.

Plainly, Hopt's reference to procedural changes reducing the "degree" of required proof meant changes in the standard that the fact finder must use to evaluate whether guilt has been established (see, e.g., De Woody v Superior Ct., 8 Cal.App.3d 52, 87 Cal.Rptr. 210; cf., State v Moyer, 387 A.2d 194 [Del]; see also, Cummings v Missouri, 4 Wall [71 US] 277). In other words, the references in those early writings to reductions in the quantum of evidence principally concerned retrospective reductions in the People's burden of proof (see, Murphy v Sowders, supra). The legislation abolishing the corroboration requirement in this case plainly does not alter the People's burden of convincing the fact finder beyond a reasonable doubt of defendant's guilt. Hence, it is not the type of procedural change that was contemplated in Hopt and Calder.

It is often argued that when the Hopt court referred to reductions in the " amount or degree" of proof ( 110 US, at 589, 590, supra [emphasis supplied]), it had two distinct concepts in mind and that, consequently, the term "amount" of proof must mean something different than merely the standard by which guilt must be proven. In this connection, reliance is often placed upon Justice Chase's inclusion in the evils to which the Ex Post Facto Clause was addressed of legislative measures that "violated the rules of evidence * * * by admitting one witness when the existing law required two" (Calder v Bull, 3 Dallas [3 US], at 389, supra). Despite its superficial appeal in cases involving corroboration rules, however, this dictum is unhelpful to defendant's position because it actually refers to an unrelated legal phenomenon that is unknown in modern American law: the "rule of number," which was directly related to the prosecuting authority's substantive burden of proof.

It seems doubtful that the Supreme Court would extend this aspect of the Calder-Hopt dictum, which concerns a particular oppressive historic practice that is long since forgotten, and use it as a basis for invalidating modern laws eliminating corroboration requirements. These corroboration rules are not the natural descendants of the rigid "rule of numbers." They have no equivalent in early common law, are principally creatures of statute and were intended simply to ensure the reliability of convictions based upon otherwise competent and legally sufficient evidence (see, People v Fuller, 50 N.Y.2d 628, 635; People v Gibson, 301 N.Y. 244, 245; Younger, The Requirement of Corroboration in Prosecutions for Sex Offenses in New York, 40 Fordham L Rev 263, 264, n 5). Moreover, extension of the seldom-invoked "procedural" aspect of the ex post facto ban to laws abrogating corroboration rules would not serve the fundamental purposes of the constitutional prohibition, as expressed in the recent Supreme Court opinions.

Defendant had ample notice that the conduct in which he engaged was prohibited and that, if convicted, he could be punished within certain legislatively prescribed guidelines. He had no legitimate right to rely on a procedural rule that affected neither his culpability nor his expected punishment but instead merely made the prosecution's case against him more difficult to prove. Indeed, as in Thompson v Missouri (supra, at 387), the repeal of the corroboration rule "did nothing more than remove an obstacle arising out of a rule of evidence". Finally, there is no claim that the legislative decision to repeal the corroboration rule applicable to defendant's case was arbitrary, vindictive or oppressive or that it was intended to single out defendant for particularly unfavorable treatment. Accordingly, the Ex Post Facto Clause has no application here, and defendant was lawfully convicted of the charged crimes even without proof to corroborate the underage victims' stories.

II. The Trial Errors

Having concluded that the evidence at defendant's trial was legally sufficient to support his convictions, we turn now to defendant's claims regarding the trial court's evidentiary rulings. Defendant contends that the trial court erred in permitting the prosecutor, over defense counsel's objection, to elicit evidence that defendant had committed similar acts of sexual abuse with Domenick M., a former pupil who had since moved away from the community. He also contends that the court's refusal to permit him to question the police investigators about their interviews with the child-witnesses was improper and prejudicial. We agree with defendant's position on both issues.

A. The Domenick M. Evidence

The evidence concerning defendant's prior misconduct with Domenick M. had no relevance to any material issue in the case and tended only to demonstrate defendant's general criminal propensity. Hence, it should have been excluded under People v Molineux ( 168 N.Y. 264) and the subsequent decisions applying the principles set forth in that case.

It is elementary that evidence of a defendant's prior criminal or immoral conduct is inadmissible if it cannot logically be linked to some specific material issue in the case (see, e.g., People v Lewis, 69 N.Y.2d 321, 325; People v Ventimiglia, 52 N.Y.2d 350, 359; People v Santarelli, 49 N.Y.2d 241; People v Allweiss, 48 N.Y.2d 40, 46-47). Although such evidence has some minimal probative worth because of its tendency to demonstrate the defendant's bad character and general criminal propensity, it is excluded as a matter of law when it has no additional relevance to a specific issue, because there is a very real danger that the trier of fact will overestimate its significance (see, People v Allweiss, 48 N.Y.2d 40, 46, supra; 1A Wigmore, Evidence § 58.2, at 1212 [Tillers rev ed]).

Admissibility of evidence under these principles is determined by reference to a two-part inquiry (People v Alvino, 71 N.Y.2d 233, 242; People v Allweiss, supra, at 46-47). The first level of this inquiry requires the proponent of the evidence, as a threshold matter, to identify some issue, other than mere criminal propensity, to which the evidence is relevant (People v Lewis, 69 N.Y.2d 321, 325, supra). Once such a showing is made, the court must go on to weigh the evidence's probative worth against its potential for mischief to determine whether it should ultimately be placed before the fact finder. This weighing process is discretionary, but the threshold problem of identifying a specific issue, other than propensity, to which the evidence pertains poses a question of law (People v Alvino, supra, at 242).

Whether prior crime evidence is actually being offered to prove propensity alone is often a subtle matter in which semantics sometimes plays an important part. For example, we recently held in People v Lewis (supra) that evidence of prior sexual contact with the same "victim," traditionally admitted under the rubric of "amorous design," was really no more than a form of propensity evidence hiding behind an assumed name and should no longer be permitted. A similar analysis applies here.

The Domenick M. testimony was offered at defendant's trial to show that the children's charges against defendant were not simply the product of rumor. The prosecution's theory was that since Domenick M. had made his accusations against defendant long before the Matthew L. story surfaced, his accusations had the ring of truth and lent credibility to the other boys' charges. However, defendant's misconduct toward Domenick M. in the past has no legitimate, legally cognizable bearing on the truthfulness of the other children's allegations. As was stated in People v Katz ( 209 N.Y. 311, 328), "[s]imple proof showing that A. shot B. at one time and place throws no light upon the charge that A. poisoned C." Similarly, proof that Domenick came forward and accused defendant of sexual misconduct before any rumors occurred and may therefore have been telling the truth may tend to support the credibility of Domenick M.'s accusation, but it throws no light upon the truthfulness of the other boys who subsequently accused defendant — except, of course, to the extent that defendant's misconduct toward Domenick M. makes it more likely that he also engaged in similar misconduct with others.

At best, this line of reasoning is a variation on the "amorous design" theme that we rejected in Lewis. In truth, it is simply another, subtler, way of saying "if defendant did it once to Domenick, he would do it again; therefore he probably abused the other children." Since this line of reasoning is nothing more than a disguised "propensity" argument, it falls within the core of what the Molineux rule prohibits and should therefore have been excluded.

Finally, the Domenick M. evidence was highly prejudicial and therefore not "harmless," despite the fact that eight other boys recounted essentially the same story. The Domenick M. evidence was qualitatively different from the testimony of the other boys, in that it was not subject to impeachment under the defense's "rumor" theory. Because of the relative strength of this evidence, particularly as it was bolstered by the grandmother's corroborative testimony, there is a serious danger that the jury used it to draw the impermissible inference ("he did it before, so he probably did it this time too") and to resolve any doubts it might otherwise have had about the other boys' stories in the People's favor. Again, it is precisely this type of result that the Molineux rule was adopted to prevent.

B. The Investigating Officers' Testimony

We also conclude that defendant was improperly denied the right to present his case by the trial court's ruling foreclosing examination of the two investigating officers about the manner in which the child-witnesses were first questioned. It is well established that the trial courts have broad discretion to keep the proceedings within manageable limits and to curtail exploration of collateral matters (see, e.g., People v Sorge, 301 N.Y. 198). However, extrinsic proof tending to establish a reason to fabricate is never collateral and may not be excluded on that ground (see, Richardson, Evidence § 503 [Prince 10th ed]; cf., People v Thomas, 46 N.Y.2d 100 [proof offered to establish motive to fabricate too remote]). Further, the trial court's discretion in this area is circumscribed by the defendant's constitutional rights to present a defense and confront his accusers (see, Davis v Alaska, 415 U.S. 308; Chambers v Mississippi, 410 U.S. 284; Washington v Texas, 388 U.S. 14; Pointer v Texas, 380 U.S. 400; People v Gissendanner, 48 N.Y.2d 543, 546).

Here, both the constitutional and the evidentiary rules were breached when the trial court precluded defense counsel from questioning the police officers in the manner proposed. Defense counsel had a good-faith basis for the proposed line of questioning, and no one challenges the bona fides of his claim, which was based upon statements in the investigators' reports (see, People v Gilliam, 37 N.Y.2d 722, revg on dissent below 45 A.D.2d 744 [Hopkins, J.]). Although the trial court concluded that the proposed line of questioning was "collateral," its conclusion is obviously flawed, since the questioning concerned more than the credibility of the People's witnesses in general and went instead to a possible reason for fabrication by these impressionable witnesses, i.e., the investigators' suggestive comments.

For that reason, the fact that defendant had a limited opportunity to explore the matter during his cross-examination of the child-witnesses does not cure the error in the court's evidentiary ruling. Indeed, the evidentiary error lies precisely in the trial court's refusal to permit extrinsic proof of a matter that went beyond a mere attack on the witnesses' general credibility.

Further, contrary to the suggestion of the dissent, the information defense counsel sought to elicit from the investigating officers was not duplicative of the admissions he elicited from some of the child-witnesses that they had changed their stories after police questioning. Both the two children who had made such admissions and the other children who had given a detailed story for the first time after police questioning were potentially subject to impeachment on the theory that their stories had been influenced by what they had heard from the investigators. Although defense counsel attempted to explore the possibility of suggestive police comments during his cross-examination of the child-witnesses, his efforts were met with flat denials. These denials might well have been materially impeached if defense counsel had been permitted to question Investigator Kintner and Officer Barnard about their comments to these impressionable young boys.

Finally, there is no theory under which the error can be considered harmless. The issue of the manner in which the police had interviewed all of the young victims and the possibility that there had been a degree of suggestiveness in their approach went to the heart of defendant's case. The limited questioning which the court permitted concerning the manner in which one of the witnesses, Kevin F., had been interviewed was no substitute for the broader inquiry that was improperly precluded. Further, because defendant was precluded from questioning the police officers about the manner in which the boys had been interviewed, his ability to develop the defense theory was seriously curtailed and the prosecutor was placed in a position to argue on summation that any defense contention concerning the victims' motive to falsify was purely speculative.

The combination of the trial court's decision to exclude the officers' testimony as collateral and its decision to permit an extended excursion into defendant's prior sexual misconduct with Domenick M. resulted in a trial that was decidedly skewed in the People's favor. On the one hand, the prosecution was permitted to introduce highly prejudicial evidence of defendant's past crime on the theory that it somehow bolstered the other child-witnesses' credibility. On the other hand, when the defense sought to introduce proof that might have had a direct impeaching impact on the same witnesses' credibility, its efforts were rebuffed.

Contrary to the dissenters' suggestion, the resulting imbalance cannot be justified on the theory that the People should be given "leeway * * * to overcome the difficulties of proof inherent in prosecutions of this nature" (dissenting opn, at 73). There will always be cases that are difficult to prove because of their peculiar facts. However, the particular difficulties of proof that inhered in this category of cases were diminished when the special corroboration requirements were repealed. Further relaxation of the rules of evidence is not warranted.

CONCLUSION

The Ex Post Facto Clause of the United States Constitution does not require that a defendant be tried under the corroboration rules that existed at the time his alleged crimes were committed. Accordingly, the amended versions of Penal Law §§ 130.16 and 260.11 were properly applied at defendant's trial, and the evidence was legally sufficient to support the jury's verdicts, notwithstanding that most of the charges were not supported by corroborative proof. We nonetheless hold that because the trial court's erroneous evidentiary rulings were highly prejudicial, the conviction that was based on those verdicts cannot be permitted to stand.

Accordingly, the order of the Appellate Division should be reversed and a new trial ordered.

TITONE, J.

Judges KAYE, ALEXANDER and HANCOCK, JR., concur with Judge TITONE; Judge ALEXANDER concurs in a separate opinion; Chief Judge WACHTLER dissents and votes to affirm in another opinion in which Judges SIMONS and BELLACOSA concur.

Order reversed, etc.