I concur in the majority’s conclusion that defendant may not be retried on the enhancement allegations that the jury decided in his favor. I dissent from the rest of the opinion, which, with complete indifference to the double jeopardy clause of the Fifth Amendment, concludes that defendant may be retried as the actual perpetrator in the killing of Karen Taylor despite a not true finding on the gun-use allegation at his first trial.
The jury’s determination that defendant was not the trigger man seems as inexplicable to me as it does to my colleagues. The perceived incorrectness of a jury verdict in favor of a criminal defendant does not, however, create a license for appellate courts to right the wrong at all costs. The theory my colleagues contrive to avoid the consequences of the jury’s explicit factual finding is not supported by the cases they rely upon, is analytically unsound and unjust and creates a constitutional problem that will prove far more troublesome than the questionable jury determination in this case.
*1333In People v. White (1986) 185 Cal.App.3d 822 [231 Cal.Rptr. 569] another division of this court unanimously held that an accused may not be retried as the actual perpetrator after a prior jury has indicated, by way of its not true findings on gun use allegations, that he did not fire the fatal shots.
The majority rejects White and reaches a contrary conclusion based on (1) its belief that the court in White failed to take into account relevant case law and improperly relied on People v. Asbury (1985) 173 Cal.App.3d 362 [218 Cal.Rptr. 902], which my colleagues think distinguishable; (2) the theory that the jury’s rejection of the gun use allegation is not legally equivalent to a determination that the accused did not personally commit the homicide because it relates solely to sentencing; and (3) the claim that the procedural requirements for the invocation of collateral estoppel have not been met since the gun use finding was not “essential” to the first conviction and the prosecution was denied an opportunity to “fully and fairly” litigate the issue of defendant’s gun use during the first trial. I will discuss each of these points in turn.
1.
In White the defendant appealed from a conviction of two counts of first degree murder. He maintained the trial court erred in permitting the prosecution to try the case on the theory that he personally shot the victims after the jury in a previous trial had determined he had not used a firearm in connection with the killings. The court agreed and reversed the judgment.
The majority strains to discredit White by questioning the relevance of People v. Asbury, supra, Upon which White relies. In Asbury the jury at the first trial found the defendant guilty of first degree murder and robbery but rejected the special circumstance that the murder had occurred in the course of the robbery. The Court of Appeal held that the subsequent felony-murder conviction was barred by the doctrine of collateral estoppel—which is included within the Fifth Amendment’s guaranty against double jeopardy (Ashe v. Swenson (1970) 397 U.S. 436, 443 [25 L.Ed.2d 469, 475, 90 S.Ct. 1189])—because “the original jury, in finding the special circumstance not true, necessarily rejected the notion that the murder occurred during the course of the robbery.” (People v. Asbury, supra, 173 Cal.App.3d at p. 365.) The majority claims that Asbury is inapposite because, despite the court’s clear statement to the contrary, the case did not genuinely raise an issue of collateral estoppel. According to the majority, the finding of the first Asbury jury that the murder had not occurred during the commission of the robbery “impliedly acquitted Asbury of the offense of first degree murder. . . in the perpetration of a robbery.” (Maj. opn., ante, p. 1320, original italics.) *1334Relying upon language in Penal Code section 1023, a statute never mentioned in the Asbury opinion, the majority concludes that “felony murder was an offense ‘of which [Asbury] might have been convicted under the accusatory pleading’ at the first trial.” (Ibid.) My colleagues’ rationale for the result in Asbury does not hold up to analysis.
As our Supreme Court has pointed out, Penal Code section 1023 relates to the doctrine of included offenses, which is a part of the constitutional guaranty against double jeopardy. (People v. Kehoe (1949) 33 Cal.2d 711, 713 [204 P.2d 321], cert. den. Kehoe v. California (1949) 338 U.S. 834 [94 L.Ed. 509, 70 S.Ct. 39].) The statute provides as follows: “When the defendant is convicted or acquitted or has been once placed in jeopardy upon an accusatory pleading, the conviction, acquittal or jeopardy is a bar to another prosecution for the offense charged in such accusatory pleading, or for an attempt to commit the same, or for an offense necessarily included therein, of which he might have been convicted under that accusatory pleading.” This language does not apply to the situation that existed in Asbury.
First of all, felony murder is not an offense “necessarily included” within the offense of first degree murder; indeed, what the majority refers to as the “offense” of “first degree felony murder” (maj. opn., ante, p. 1321) does not exist; felony-murder is merely one of several definitions of murder in the first degree. (Pen. Code, § 189.) Ordinarily, a jury is not required to determine whether a defendant charged with first degree murder committed the offense in the course of committing any of the felonies enumerated in Penal Code section 189 unless the People prosecute the case on such a theory, in which case the jury receives CALJIC No. 8.21, which instructs that “[t]he unlawful killing of a human being ... as a result of the commission of [a specified felony] ... is murder of the first degree.” Because a felony-murder theory was not originally advanced in Asbury this instruction was not given. The factual question whether the homicide was committed in the course of another felony was presented to the jury only because of the special circumstance allegation. If, contrary to the doctrine of included offenses set forth in Penal Code section 1023, conviction of first degree murder can result in an “implied acquittal” of that same offense on a prosecution theory which, though it was not actually advanced, can be shown to have been rejected by the jury due to a finding made in connection with a penalty enhancement, then the defendants in White and this case were also impliedly acquitted of first degree murder on the theory that they directly committed the act constituting the offense and were therefore perpetrators within the meaning of the Penal Code. (Pen. Code, § 31.) For purposes of the theory of implied acquittal conjured by the majority, the facts of Asbury, White and the present case are materially indistinguishable.
*1335Moreover, if the majority’s novel theory were accepted, the instant case would present an even stronger case than Asbury for implied acquittal. As indicated, the prosecution did not initially advance a felony-murder theory in Asbury and the jury addressed the factual question only because of allegations made in order to enhance penalty. In this case, on the other hand, the prosecution did advance an aiding and abetting theory that necessitated the giving of instructions (CALJIC Nos. 3.00 and 3.01) describing the difference between a perpetrator and an aider and abettor in connection with the charged homicide. Therefore, while the Asbury jury almost certainly did not understand the relationship between the facts set forth in the enhancement allegation and those relating to the murder charge (because it was never told about the felony-murder rule), the jury in this case almost certainly did understand that relationship, making it easier here to imply acquittal on the theory defendant directly committed the homicide than it was in Asbury to imply acquittal on a felony-murder theory. Looking at it from the point of view of the majority’s distorted notion of the doctrine of included offenses, the prosecution’s theory that the defendant in this case directly committed the homicide was much more clearly “included” in the charges against him than was felony-murder “included” in the charges against Asbury.
Though the majority’s concept of included offenses and the collateral theory of implied acquittal, if consistently applied, would justify the result in this case that I think is constitutionally compelled, the majority’s analysis is fundamentally flawed. The theory of implied acquittal embraced in Penal Code section 1023 is correctly applied only in relation to a charged or necessarily included offense; it therefore has no application either in Asbury or in this case.
Asbury was correctly decided for the reason stated in that opinion—collateral estoppel, not because of an implied acquittal. The situation in that case, like the situation in White and that here, satisfies all three requirements of collateral estoppel. The party against whom the estoppel was asserted was a party at the earlier trial, that proceeding resulted in a final judgment on the merits with respect to the ultimate fact in issue,1 and, most *1336important for present purposes, the issue necessarily decided at the previous trial was identical to the one sought to be relitigated. The Asbury court found that when the first jury refused to find that the defendant committed the murder “in the commission of’ the robbery (the phrase used in the special circumstance instruction), it was effectively declaring that he did not commit the murder “as a result of’ the robbery (the language of the felony-murder instruction). The court was in effect saying that the finding of the jury at the previous trial was not simply an evidentiary determination but an “ultimate fact” (i.e., a finding involving the application of law to fact) with respect to both the special circumstances allegation and the question of felony murder. Asbury was thus a classic case for application of the rule of collateral estoppel because, as the Supreme Court has stated, “when an issue of ultimate fact has once been determined by a valid and final judgment, that issue cannot again be litigated between the same parties in any future lawsuit.” (Ashe v. Swenson, supra, 397 U.S. 443 [25 L.Ed.2d at p. 475]; People v. Taylor (1974) 12 Cal.3d 686, 693 [117 Cal.Rptr. 70, 527 P.2d 622]; Rest.2d Judgments, § 27.)
Similarly, when the jury in the present case rejected the allegation that Pettaway “personally used a firearm during the commission of [the alleged homicide]” (the phrase used in the gun use instruction, CALJIC No. 17.19), it was effectively declaring that he did not “directly and actively commit the act constituting the crimes” (the language of CALJIC No. 3.00, given as part of the aiding and abetting instruction), thereby negating an ultimate fact necessary to convict him on the ground that he actually perpetrated the homicide. Jeopardy therefore attached in this case in precisely the same manner it did in Asbury. Just as Asbury might have been convicted on a felony-murder theory under the accusatory pleading at the first trial, so too at his first trial might Pettaway have been convicted of murder as a perpetrator, as the district attorney vigorously urged.
The special circumstances allegation in Asbury, like the gun-use allegation in White and in this case, was not included in the charges in order to establish an element of any offense, but simply to enhance the penalty. (Pen. Code, § 190.2.) The Asbury court recognized, however, that this was not crucial. With “realism and rationality,” the court simply inquired “whether a rational jury could have grounded its verdict upon an issue other than that which the defendant seeks to foreclose from consideration” (Ashe v. Swenson, supra, 397 U.S. 436, 444 [25 L.Ed.2d at pp. 475-476], fn. omitted) and found that it could not. This is what we should do, rather than employ “the hypertechnical and archaic approach of a 19th century pleading book.” (Ibid. [25 L.Ed.2d at p. 475].)
*1337In a later part of its opinion the majority also takes issue with the White court’s refusal to accept the argument that Penal Code section 954 (which, as pertinent, provides that “An acquittal on one or more counts shall not be deemed an acquittal of any other count”) permitted retrial of the gun use issue. (Maj. opn., ante, p. 1329.) The majority’s point is summed up in the following three sentences: “Had Pettaway contended in his [initial] appeal that his conviction for murder must be reversed on the grounds it was inconsistent with the finding on the enhancement, he could not have prevailed under section 954 and [People v. Amick (1942) 20 Cal.2d 247 [125 P.2d 25] and People v. Federico (1981) 127 Cal.App.3d 20 [179 Cal.Rptr. 315]]. There is no logical reason why the result should differ on retrial after his murder conviction was reversed for Beeman error. The finding on the enhancement simply has no effect on the murder charge in the first or the second trial.” (Maj. opn., ante, p. 1330.)
This reasoning reveals the majority’s failure to understand the difference between an inconsistent verdict at a single trial and multiple trials of the same person for the same offense. In People v. Tideman (1962) 57 Cal.2d 574 [21 Cal.Rptr. 207, 370 P.2d 1007], the Supreme Court described the statutory history of 954, stating that the legislative purpose of the language here relevant was to make it “altogether clear that under current criminal procedure . . . ‘[t]he doctrine of double jeopardy has no application [in a single criminal action] to a defendant who is tried but once on several counts. [Fn. omitted.]’ ” (Id., at p. 581, quoting People v. Chessman (1951) 38 Cal.2d 166, 193 [238 P.2d 1001], cert den. Chessman v. California (1952) 343 U.S. 915 [96 L.Ed. 1330, 72 S.Ct. 650].) The reasons the law tolerates inconsistent verdicts, which were set forth in White,2 are entirely unrelated *1338to the fundamentally different considerations that come into play when an accused person is subjected to more than one trial for the same offense.
My colleagues say they can think of no “logical reason” why the rationale of Penal Code section 954 should not apply where an individual is subjected to a second trial for the same offense. The “logical reason” they cannot fathom is the very principle that informs the prohibition against double jeopardy. “The underlying idea, one that is deeply ingrained in at least the Anglo-American system of jurisprudence, is that the State with all its resources and power should not be allowed to make repeated attempts to convict an individual for an alleged offense, thereby subjecting him to embarrassment, expense and ordeal and compelling him to live in a continuing state of anxiety and insecurity, as well as enhancing the possibility that even though innocent he may be found guilty.” (Green v. United States (1957) 355 U.S. 184, 187-188 [2 L.Ed.2d 199, 204, 78 S.Ct. 221, 61 A.L.R.2d 1119].) This ancient idea, whose origins can be traced to Greek and Roman times and which became established in the common law of England long before our Nation declared its independence (Benton v. Maryland (1969) 395 U.S. 784, 795 [23 L.Ed.2d 707, 716, 89 S.Ct. 2056]), is too firmly rooted in our criminal jurisprudence to permit question. As Chief Justice Burger once observed, “where the Double Jeopardy Clause is applicable, its sweep is absolute. There are no ‘equities’ to be balanced, for the Clause has declared a constitutional policy, based on grounds which are not open to judicial examination.” (Burks v. United States (1978) 437 U.S. 1, 11, fn. 6 [57 L.Ed.2d 1, 9, 98 S.Ct. 2141], italics added.)
The White court was indubitably correct in concluding that the prohibition of double jeopardy cannot be compromised by application of Penal Code section 954 or its rationale.
Having cavalierly disposed of White, the majority relies on People v. Lopez (1982) 131 Cal.App.3d 565 [182 Cal.Rptr. 563] and People v. Nunez (1986) 183 Cal.App.3d 214 [228 Cal.Rptr. 64] (rev. den. Oct. 16, 1986) as support for its position. In Lopez, the defendant was found guilty on six counts of assault with a deadly weapon but the jury found he did not personally use a firearm. On appeal he claimed there was insufficient evi- . dence to support the verdict and asserted that in light of the jury’s finding on the use allegation the court was bound to assume he was convicted as an aider and abettor and not a principal. The court found the case analogous to those involving inconsistent verdicts3 and rejected the defendant’s argu*1339ment, finding that the inconsistency between the verdict on each of the offenses and the finding on the enhancement did not invalidate the convictions. Lopez is obviously different from the instant case, which does not involve rationally inconsistent verdicts or findings.4 More importantly, since Lopez had been tried only once his case has nothing to do with the principles of collateral estoppel and double jeopardy that are directly implicated in this case, where defendant appears destined to face retrial for the same offense on a factual theory explicitly rejected by a prior jury.
My colleagues’ reliance upon Lopez is curious, because the opinion in that case does not support but repudiates their principal assertion that a factual finding on an alleged enhancement is for double jeopardy purposes significantly different from a factual determination relating to an element of an offense. When it commenced its inquiry into the legal effect of the inconsistency between the verdict on each of the offenses and the finding that he did not use a firearm, the court in Lopez considered whether to ignore the finding of no use on the ground that, technically, an enhancement may be differentiated from an element of the offense—which is the course my colleagues take. The court refused to take this position because it realized that for the purposes of its legal analysis a negative factual finding regarding an enhancement cannot rationally be distinguished from a factual finding regarding an element of an offense. As stated in Lopez, “although the amended information alleges ‘enhancement’ rather than an ‘offense’ . . ., the degree of factual inconsistency is no greater, or no less, than in the cases discussed above [in which the inconsistency was solely between offenses].” (Lopez, supra, 131 Cal.App.3d at p. 570.)
The majority also relies on People v. Nunez, supra, 183 Cal.App.3d 214, where the defendant was convicted of conspiracy to commit murder for financial gain and first degree murder, although the jury found he did not personally use a firearm. Nunez maintained that since the jury had returned a not true finding on the gun-use allegation he had been convicted as an aider and abettor and, therefore, should benefit from the juries’ conclusions in his coconspirators’ cases that the killing was not committed with malice *1340for financial gain.5 The court properly rejected these claims. First, it found that Nunez’s motive for participating in the killing was not necessarily derived from his coconspirators and that he could be convicted of killing for financial gain despite the contrary finding at the trial of one of his coconspirators.
The court further reasoned that the verdict in the trial of Nunez’s coconspirator, Medina, did not necessarily establish that Medina had been convicted as the perpetrator or that the jury had determined he had acted without malice. Because the record was insufficient to determine whether Medina had been tried solely as the perpetrator the court concluded that the requirements for applying collateral estoppel had not been met: “we cannot say the previous jury necessarily determined the only principal between the two men did not harbor malice. This issue not being decided adversely to the prosecution, collateral estoppel would not bar defendant’s conviction of the crime of murder on the theory that he was either the direct perpetrator or guilty as an aider and abettor.” (Nunez, supra, 183 Cal.App.3d at p. 227.) Thus, the court relied heavily on “the difficulty of identifying the issues resolved against the People in the prior trial.” (Id., at p. 221, relying on People v. Taylor (1974) 12 Cal.3d 686, 696 [117 Cal.Rptr. 70, 527 P.2d 622].) There is no such difficulty in the present case, where we know exactly what issue was resolved against the People at the first trial.
I have no quarrel with the reasoning in Nunez. However, in reaching its conclusion the court considered the effect of a negative finding on a gun-use allegation and made the bewildering statement—seized upon by the majority here—that “[a] jury’s finding on an alleged enhancement that an accused was not armed with a firearm or did not personally use a firearm does not necessarily mean that the accused was not a direct perpetrator of the crime.” (People v. Nunez, supra, 183 Cal.App.3d at pp. 225-226.) Taken out of context, this statement is irrational. A jury which finds that a defendant did not personally use a handgun in a case in which the victim died from gunshot wounds is clearly stating its belief that the defendant was not the perpetrator.
This is not what the Nunez court either needed or could have meant to say. The statement appears in Nunez in the context of a discussion of People v. Lopez, supra, 131 Cal.App.3d 565, in which, as I have described, the court held that inconsistencies between a verdict on an offense and a finding *1341on an enhancement did not invalidate the verdict on the offense. Therefore, the idea the Nunez court almost certainly intended to convey by the statement in question is that a jury’s finding on an alleged enhancement that an accused did not personally use a firearm does not necessarily mean that the accused could not be convicted as a perpetrator pursuant to an inconsistent verdict on the offense in the same case.
It also bears pointing out that the Nunez court accepted the conclusion in Lopez that an inconsistent jury finding cannot be ignored simply because it relates to an enhancement rather than an offense. (Nunez, supra, 183 Cal.App.3d at p. 226.) Thus, to the dubious extent that Nunez and Lopez, which did not involve multiple trials of the same person for the same offense, are at all germane to the issues before us here, they reject rather than support the basic idea upon which my colleagues construct their house of cards.
2.
The second theory advanced by the majority attempts to distinguish the jury’s negative finding on the gun use allegation from a determination that the accused was not the actual killer. The majority argues that the jury’s finding on the gun-use allegation ought not preclude appellant’s retrial as the actual perpetrator because “[w]e are not concerned with the subject of punishment following conviction. A use enhancement is relevant only to punishment.” (Maj. opn., ante, p. 1325.) To be sure, an affirmative personal use finding would technically affect only the length of sentence, and not the issue of guilt on the substantive offense (though the jury is unaware of the limited effect of its determination). However, we do not have an affirmative gun-use finding, but a negative finding, which has implications that go beyond sentencing. As I have been insisting, the plain and unassailable reality is that when in a murder case the jury determines that a gun-use allegation is untrue it is unmistakably declaring that the defendant was not the perpetrator of the homicide. That this factual determination addresses an allegation pled separately from the offense is not only unknown to the jury but wholly beside the legal point. As earlier explained, the negative finding on the gun use relates to an issue of “ultimate fact.” Once such an issue has been validly determined in favor of a defendant, as it has in this case, it cannot be litigated between the same parties in any future lawsuit. (Ashe v. Swenson, supra, 397 U.S. 436, 443 [25 L.Ed.2d 469, 475].)
The principle that the People are precluded from relitigating the same issue against the same defendant based on the same factual circumstances may apply even if the original determination was sought in connection with the issue of penalty, did not formally result in an acquittal on a charged offense, and did not necessarily negate the possibility of prosecution for an *1342independent offense. When in Arizona v. Rumsey (1984) 467 U.S. 203 [81 L.Ed.2d 164, 104 S.Ct. 2305] and Bullington v. Missouri (1981) 451 U.S. 430 [68 L.Ed.2d 270, 101 S.Ct. 1852], the United States Supreme Court accorded double jeopardy protection to special verdicts rendered by fact finders refusing to impose the death penalty, it reasoned that the verdicts were essentially indistinguishable from acquittals on charged offenses because (1) the prosecution had the burden of proving statutorily defined facts beyond a reasonable doubt; (2) the court or jury was required to make specific findings or render a special verdict; (3) the factfinder’s decision was based on a determination that the prosecution either had or had not proved its case; and (4) the determination was made following a hearing which involved the submission of evidence and presentation of argument. (Arizona v. Rumsey, supra, 467 U.S. at pp. 209-211 [81 L.Ed.2d at pp. 170-171]; Bullington v. Missouri, supra, 451 U.S. at p.438 [68 L.Ed.2d at pp. 278-279].) All of these requirements were met in the present case with respect to the gun-use allegation. Thus, the fact that the allegation related, inter alia, to sentencing does not in and of itself permit the substance of the allegation to be relitigated.
My colleagues take me to task for allegedly ignoring the reluctance of the Supreme Court in Rumsey and Bullington to extend the double jeopardy principle to sentencing and the limited rationale which supports the holdings in those cases. (Maj. opn., ante, p. 1328.) The judicial reluctance to which my colleagues refer clearly does not relate to the type of situation presented in this case. Rumsey and Bullington both involved capital murder charges under state laws that separated trial on the issue of guilt from that on the issue of penalty. In both the triers of fact found against the defendant on the question of guilt but, at the sentencing trial, found that the penalty of death should not be imposed. It was the latter findings that barred the state from seeking to reimpose the death penalty at a second trial. The Bullington court noted that the principle of double jeopardy ordinarily does not apply to sentencing because “[t]he imposition of a particular sentence usually is not regarded as an ‘acquittal’ of any more severe sentence that could have been imposed.” (Bullington v. Missouri, supra, 451 U.S. at p.438 [68 L.Ed.2d at p. 278].) Double jeopardy was nevertheless applied in Bullington and Rumsey because in those cases the prosecution had the burden of proving certain facts beyond a reasonable doubt and the sentencing hearing “resembled and, indeed, in all relevant respects was like the immediately preceding trial on the issue of guilt or innocence.” (Ibid. [68 L.Ed.2d at p. 279], Arizona v. Rumsey, supra, 467 U.S. at pp. 209-210 [81 L.Ed.2d at pp. 170-171].) The doubts expressed by the Supreme Court about the application of double jeopardy principles to matters that ordinarily relate only to sentencing certainly do not apply to the situation presented in this case, *1343where the critical fact was established at a proceeding that did not merely resemble but was the trial on the issue of guilt or innocence.
The majority’s suggestion that, out of mercy or a desire for leniency, the jury might have issued a negative finding on the gun use even if it really believed appellant was the trigger man is not only inconceivable as a matter of fact but highly irregular as a matter of law. The notion that the jury may have acted out of mercy or a desire for leniency is not based upon anything in the record, but was presumably suggested by judicial reliance on such a possibility in cases like People v. Nunez, supra, 183 Cal.App.3d at p.226, involving inconsistent verdicts at a single prosecution. (E.g., Dunn v. United States (1932) 284 U.S. 390, 393 [76 L.Ed. 356, 359, 52 S.Ct. 189, 80 A.L.R. 161], quoting Steckler v. United States (2d Cir. 1925) 7 F.2d 59, 60; People v. Amick, supra, 20 Cal.2d 247, 252, quoting People v. Horowitz (1933) 131 Cal.App.Supp. 791, 793-794 [19 P.2d 874]; see also Standefer v. United States (1980) 447 U.S. 10, 11 [64 L.Ed.2d 689, 692-693, 100 S.Ct. 1999].) I am aware of no authority for the proposition that a person may be twice prosecuted for the same offense because an ultimate fact decided adversely to the People ip the first trial may have resulted from undeserved mercy or a desire for leniency.
Absent specific and strong reason to think otherwise, an appellate court is required to assume that a jury verdict in a case under review resulted from a proper understanding and application of the trial court’s instructions (Zuckerman v. Underwriters at Lloyd’s (1954) 42 Cal.2d 460, 478-479 [267 P.2d 777]; Palmer v. Ted Stevens Honda, Inc. (1987) 193 Cal.App.3d 530, 536 [238 Cal.Rptr. 363]; Trapani v. Holzer (1958) 158 Cal.App.2d 1, 6 [321 P.2d 803]), which in a criminal case includes the admonition that the subject of punishment is not to be discussed or considered by the jury and must not in any way affect its verdict. (CALJIC No. 17.42.) We must presume that the jurors were intelligent persons who understood and followed the instructions they received. (People v. Romo (1975) 47 Cal.App.3d 976, 990 [121 Cal.Rptr. 684]; People v. Powell (1960) 186 Cal.App.2d 54, 59 [8 Cal.Rptr. 707].) If reviewing courts are permitted to attribute an unapparent but congenial meaning to a jury verdict with which they otherwise do not agree, they will have freed themselves from the constraints of neutral legal principles and made a mockery of the appellate process.
The record presents absolutely no reason to believe that, as the majority wildly speculates, the jury rejected the gun use allegation out of mercy or “through confusion or ennui.” (Maj. opn., ante, p. 1324.) On the contrary, what evidence we do have of the jury’s motives indicates exactly the opposite of that which my colleagues conveniently imagine. As the majority acknowledges, on the third day of deliberations the jury conveyed a note to *1344the court stating: “please explain the law concerning complicity, for example, in this case may the defendant be convicted of murder or attempted murder without having personally fired the bulletsT’ (Maj. opn., ante, p. 1315, italics added.) Shortly after receiving the requested aiding and abetting instructions the jury returned verdicts of guilty for both first degree murder and attempted murder. Significantly, the jury found the gun-use allegations true as to the attempted murder but untrue as to the murder. The nature of the legal question the jury put to the judge shortly before it returned its verdict and the different findings on the two gun-use allegations plainly show the jury was discriminating in its evaluation of the evidence and that it did not believe defendant pulled the trigger of the gun that killed Karen Taylor.
The record shows not only that the jury knew what it was doing and meant what it said but that the district attorney who prosecuted the case actually anticipated this result.
It was defense counsel—clearly fearful of a compromise verdict—who objected to the giving of aiding and abetting instructions at the close of trial. When the jury later requested such instructions the district attorney urged that they be given and stated that his failure to earlier insist that this be done was inadvertent. The prosecutor’s rationale for giving the aiding and abetting instructions was as follows: “What I neglected to bring to the Court’s attention and to the jury’s attention is because the defense in this case was that the defendant did not commit either offense, that if the jury in fact rejected the defense’s version of what happened and accepted the prosecution’s version, and the evidence I think was much stronger on the attempted murder as point out to the defendant [s/c] than on the murder charge, that if the jury did in fact accept the defendant as being the principal on the attempted murder, and having rejected anyone else’s presence, then only two people could have been involved in the murder: The defendant or Ms. Lawanna Walker. And the jury could still convict the defendant of murder, if they believed Lawanna Walker had in fact fired the fatal shots . . . .” (Italics added.)
The result foreseen by the district attorney is the result that came to pass. He must now live with it, because, as apparently still needs to be said, the constitutions of this state and nation forbid placing a person twice in jeopardy for the same criminal act. (Cal. Const., art. I, § 15; U.S. Const., 5th Amend.)
3.
Finally, the majority claims collateral estoppel cannot be applied in this case since (1) the gun-use finding was not “essential” to the judgment; and *1345(2) the prosecution was not permitted to fully litigate the question of appellant’s actions at the first trial due to Ms. Walker’s refusal to testify.
The majority focuses on language from Newton v. Superior Court (9th Cir. 1986) 803 F.2d 1051, 1057, cert. den. (1987) 481 U.S. 1070 [95 L.Ed.2d 873, 107 S.Ct. 2464], which indicates that collateral estoppel will only apply to those issues that were actually litigated and “ ‘essential to the judgment’ ” and argues that collateral estoppel should not apply in this case because the use finding is “adjunctive” in nature. This is casuistry. The principle that collateral estoppel only applies to issues essential to the judgment ensures that parties are precluded from relitigating only those issues actually decided by the trier of fact. In the absence of special findings, a later court may know that a subsidiary issue was resolved only if it was “essential to the judgment.” It is elementary, however, that inquiry whether a particular issue was essential to an earlier judgment is necessary only if that judgment does not explicitly address the issue or the question is otherwise in doubt. Such inquiry is obviously unnecessary where, as in this case, there can be no doubt either that the original finder of fact specifically addressed and decided the issue in question (because it did so in a special verdict) or that the finding was essential (because, as required by law, the court directed that the finding be made and included it in the judgment). Nor can there be any doubt that the issue originally decided—whether appellant pulled the trigger of the gun that killed Karen Taylor—is the precise issue that would be presented again at the second trial. As pointed out in the Restatement, “[w]hen an issue is properly raised, by the pleadings or otherwise, and is submitted for determination, and is determined, the issue is actually litigated within the meaning of the [rule of collateral estoppel].” (Rest.2d Judgments, supra, § 27(d), p. 255.)
It bears mentioning, in this connection, that double jeopardy protection is regularly accorded criminal defendants in instances in which the jury was much less explicit than the jury here. Under the theory of implied acquittal codified in Penal Code section 1023, “a verdict of guilty of a lesser included offense constitutes an implied acquittal of the greater offense of which the jury could have convicted the defendant.” (Stone v. Superior Court (1982) 31 Cal.3d 503, 511 [183 Cal.Rptr. 647, 646 P.2d 809], fn. omitted; Green v. Superior Court, supra, 355 U.S. 184.) Further, “[i]n some circumstances, double jeopardy bars a retrial even though no verdict has been rendered. Once jeopardy attached, discharge of the jury without a verdict is tantamount to an acquittal and prevents a retrial, unless the defendant consented to the discharge or legal necessity required it. [Citations.]” (Stone, supra, at p. 516.) It seems to me anomalous to grant double jeopardy protection where a jury has not resolved an issue at all, or done so only by implication, *1346but to deny protection where, as in this case, the jury has explicitly ruled in favor of the defendant regarding an ultimate fact.
Perhaps the most astonishing aspect of the majority opinion is the notion that collateral estoppel should not be applied in this case because the prosecution did not at the first trial have a full and fair opportunity to litigate the issue of defendant’s guilt. (Maj. opn., ante, pp. 1325-1326.) The majority bases this theory on the analysis employed by the United States Supreme Court in Standefer v. United States, supra, 447 U.S. 10, an opinion which has almost nothing to do with the issue before us here. In Standefer the defendant was indicted, inter alia, for aiding and abetting a named Internal Revenue Service agent in accepting unlawful compensation. Prior to the indictment, the IRS agent was acquitted of certain of the violations the defendant was accused of aiding and abetting. The defendant thereupon moved to dismiss his indictment as to these violations on the ground that since the agent had been acquitted of such violations, the defendant could not be convicted of aiding and abetting them. In other words, unlike the present case, the defendant in Standefer was not subject to retrial; there was no prior jury finding that he could not have committed the offense in the manner charged.
In affirming the trial court’s denial of the motion to dismiss, the Supreme Court expounded at some length upon the doctrine of nonmutual collateral estoppel in the context of a criminal case. In explaining that the prosecution is often without the kind of “ ‘full and fair opportunity to litigate’ ” that is a prerequisite of estoppel, the court specified the aspects of the criminal law that make this so: “the prosecution’s discovery rights in criminal cases are limited, . . .; it is prohibited from being granted a directed verdict or from obtaining a judgment notwithstanding the verdict no matter how clear the evidence of guilt . . . ; it cannot secure a new trial on the ground that an acquittal was plainly contrary to the weight of the evidence . . . ; and it cannot secure appellate review where a defendant has been acquitted.” (Standefer v. United States, supra, 447 U.S. at p. 22 [64 L.Ed.2d at p. 699].) Moreover, the court observed, “[t]he application of nonmutual estoppel in criminal cases is also complicated by the existence of rules of evidence and exclusion unique to our criminal law.” (Id., at p. 23 [64 L.Ed.2d at p. 700].) Thus, “[i]t is frequently true in criminal cases that evidence inadmissible against one defendant is admissible against another. The exclusionary rule, for example, may bar the government from introducing evidence against one defendant because that evidence was obtained in violation of his constitutional rights. ... In such circumstances, where evidentiary rules prevent the Government from presenting all its proof in the first case, application of nonmutual estoppel would be plainly unwarranted.” (Id., at pp. 23-24 [64 L.Ed.2d at p. 700], fn. omitted.)
*1347The reason the court refused to estop the initial prosecution of one person because of the acquittal of another on related charges is because of the unfairness that would result if the present defendant were permitted to vicariously reap the benefit of rules preventing the prosecution from fully presenting its proof against the person first prosecuted and from appealing the acquittal of that other person. No such unfairness would flow from the application of estoppel in this case. The prosecution’s discovery rights against defendant were no different at his first trial than they would be at another. Nor was any rule of evidence applicable to the first trial that would not apply at a second. Nor does the absence of any remedial procedure at defendant’s first trial provide any equitable justification for a second prosecution.
The reasons the majority thinks it would be “unfair” to estop a second prosecution in this case are not among those referred to in Standefer, nor so far as I am aware have they ever previously been relied upon by any court to justify this result. The majority would permit a second prosecution of defendant as perpetrator simply because they agree with the prosecution that he was the perpetrator and because the district attorney at the first trial assertedly was prevented from making a fair presentation of admissible evidence “because of an artificial curtailment of the People’s theory and proof . . . . ” (Maj. opn., ante, p. 1326.)
The “artificial curtailment” of the People’s proof consisted of the exercise of the Fifth Amendment right to remain silent by Lawanna Walker, who is now willing to waive that right, and is expected to testify that defendant was the trigger man. This change of heart appears to be the result of a favorable plea bargain which, though it could have been, was apparently not offered Ms. Walker at the first trial. The majority seems therefore to be saying that a district attorney may withhold immunity, a favorable plea bargain or some other advantage sought by a witness who otherwise refuses to testify and hope that the jury will convict without that testimony; if the district attorney is proved wrong and the jury acquits, the prosecutor can then simply grant the witness the desired favor in return for the necessary testimony and retry the defendant for the same offense without violence to the doctrine of collateral estoppel. Such a scenario, which permits a district attorney to profit from his own miscalculation, should be inconceivable. Collateral estoppel must be applied independently of a prosecutor’s strategy decisions. The district attorney in this case, who, as earlier pointed out, anticipated that the jury might find defendant guilty as an aider and abettor, could as easily before the first trial as afterwards have made Ms. Walker the *1348offer that appears to have persuaded her to agree to testify. To use the witness’s belated change of heart—or, more likely, the district attorney’s belated exercise of his considerable power to induce such a change of heart—as a basis upon which to permit defendant to be retried for the same offense, and to do this in the name of fairness, gives a new and grotesque meaning to the word “unfair,” which is certainly not the one the Standefer court had in mind.
Defendant’s guilt or innocence of the offense for which the district attorney would retry him is irrelevant. As has been pointed out, “[w]hile the double jeopardy doctrine is not designed to protect criminals, it must be applied even though the result is frustration in whole or in part of punishment for crime in a particular case, and the fact that the result of upholding a plea of former jeopardy will be that [an] accused will go without sentence or punishment for an offense to which he had pleaded guilty does not alter his rights or change his position as to his right to plead former jeopardy.” (22 C.J.S., Criminal Law, § 238, p. 617, citations omitted.)
The majority attempts to show that this case does not involve a double jeopardy issue by theorizing that defendant is subject to “continuing jeopardy,” as that concept is explained in Justices of Boston Municipal Court v. Lydon (1984) 466 U.S. 294 [80 L.Ed.2d 311, 104 S.Ct. 1805]. My colleagues have again missed the point. The concept of continuing jeopardy, which simply permits the retrial of a defendant whose conviction was reversed on appeal, clearly does not permit the retrial of an offense, such as that involved in this case, where “there has been some event, such as an acquittal, which terminates the original jeopardy.” (Richardson v. United States (1984) 468 U.S. 317, 325 [82 L.Ed.2d 242, 251, 104 S.Ct. 3081], italics added; see discussion, ante, fn. 1, p. 1335.) Not one of the federal and state cases cited by the majority involved an attempt to relitigate an ultimate fact previously determined in favor of the defendant. The cases are all inapposite.
United States v. Ball (1896) 163 U.S. 662 [41 L.Ed. 300, 16 S.Ct. 1192], one of the early cases on the subject, illustrates very well why a person in the position of the defendant in the present case is not subject to continuing jeopardy. Ball involved murder charges against three defendants who were tried together. The jury acquitted one and convicted the other two. On the initial appeal the two convictions were reversed because the indictment, “by reason of failing to aver either the time or the place of the death of [the victim], was fatally defective, and would not support a sentence for murder . . . .” (Id., at p.664 [41 L.Ed. at p. 301].) The trial court thereupon *1349dismissed the defective indictment and returned a new indictment against all three defendants. The two defendants who had been convicted then “filed a plea of former jeopardy, by reason of their trial and conviction upon the former indictment, and of the dismissal of that indictment.” (Id., at p. 665 [41 L.Ed. at p. 301].) The remaining defendant, Millard Fillmore Ball, filed a similar plea but relied primarily on the fact of his acquittal at the first trial. The state appellate court denied all three pleas on the ground that the original indictment was insufficient to serve as an indictment for murder and that the three defendants were therefore not initially placed in jeopardy. All three defendants were convicted at the second trial. The United States Supreme Court approved the reprosecution and affirmed the convictions as to the two defendants whose initial convictions earlier had been reversed, holding that “a defendant, who procures a judgment against him on an indictment to be set aside, may be tried anew upon the same indictment, or upon another indictment, for the same offence for which he had been convicted. [Citations.]” (Id., at p. 672 [41 L.Ed. at p. 303].) In other words, the jeopardy of the two defendants who were initially convicted never terminated and they were subject to “continuing jeopardy.”
It is in connection with the third defendant, however, that Ball is, for present purposes, most illuminating. The court reversed the conviction of this defendant, concluding that “a general verdict of acquittal upon the issue of not guilty to an indictment undertaking to charge murder, and not objected to before the verdict as insufficient in that respect, is a bar to a second indictment for the same killing.” (United States v. Ball, supra, at p. 669 [41 L.Ed. at p. 302].) In reaching this result, the court essentially adopted language in the dissenting opinion of Justice Livingston in People v. Barrett (N.Y. 1803) 1 Johns. 66, which until Ball had been the leading American case on application of the double jeopardy clause. Justice Livingston’s views are particularly relevant to the case before us. “ ‘This case,’ ” he stated, “ ‘presents the novel and unheard of spectacle, of a public officer, whose business it was to frame a correct bill, openly alleging his own inaccuracy or neglect, as a reason for a second trial, when it is not pretended that the merits were not fairly in issue on the first. That a party shall be deprived of the benefit of an acquittal by a jury, on a suggestion of this kind, coming too from the officer who drew the indictment, seems not to comport with that universal and humane principle of criminal law, “that no man shall be brought into danger more than once for the same offence.” It is very like permitting a party to take advantage of his own wrong. If this practice be tolerated, when are trials of the accused to end? . . . [T]he prosecutor, if he be dissatisfied and bent on conviction, has nothing to do but to tell the court that his own indictment was good for nothing; that it has no venue, or is deficient in other particulars, and that, therefore, he has *1350a right to a second chance of convicting the prisoner, and so on, toties quoties.' ” (United States v. Ball, supra, at pp. 667-668 [41 L.Ed. at p. 302], quoting People v. Barrett, supra, 1 Johns, at p. 74, dis. opn. of Livingston, J.)
The majority would permit the previously litigated question whether defendant directly committed the murder of Karen Taylor to be relitigated due in part to the failure of the prosecution at the first trial to obtain the more convincing testimony that it has apparently now secured. This reasoning is constitutionally proscribed. “The Double Jeopardy Clause forbids a second trial for the purpose of affording the prosecution another opportunity to supply evidence which it failed to muster in the first proceeding. This is central to the prohibition against successive trials.” (Burks v. United States, supra, 437 U.S. 1, 11 [57 L.Ed.2d at p. 9], fn. omitted; see also, Greene v. Massey (1978) 437 U.S. 19 [57 L.Ed.2d 15, 98 S.Ct. 2151]; Hudson v. Louisiana (1981) 450 U.S. 40 [67 L.Ed.2d 30, 101 S.Ct. 970].)
As indicated at the outset, I agree that the evidence suggests defendant may have been the actual killer and not merely an aider and abettor. However, under our system it is not the appellate courts but the jury that is the ultimate arbiter of the facts; its determination should not be disregarded simply because other minds would have reached a different conclusion or to provide the losing side an opportunity to present a stronger case.
Defendant has been subjected to a murder trial at which the jury’s expressly determined he did not personally shoot Karen Taylor. As that issue has been clearly decided adversely to the prosecution it cannot be relitigated without offending the Constitutions of California and the United States. Accordingly, I would affirm the judgment.
On January 26, 1989, the concurring and dissenting opinion was modified to read as printed above. A petition for a rehearing was denied January 26, 1989. Kline, P. J., was of the opinion that the petition should be granted. Respondent’s petition for review by the Supreme Court was denied April 20, 1989. Mosk, J., and Broussard, J., were of the opinion that the petition should be granted.
Rejection of the special circumstances allegation in Asbury, like the findings of no gun use in White and the present case, could not have been appealed and therefore constituted a judgment which, for purposes of collateral estoppel, was both valid and final. As has authoritatively been stated, “[f]or purposes of issue preclusion .... ‘final judgment* includes any pri- or adjudication of an issue in another action that is determined to be sufficiently firm to be accorded conclusive effect.” (Rest.2d Judgments, § 13.) “To be ‘final’ for purposes of collateral estoppel the decision need only be immune, as a practical matter, to reversal or amendment.” (Miller Brewing Co. v. Jos. Schlitz Brewing Co. (7th Cir. 1979) 605 F.2d 990, 996, cert. den., 444 U.S. 1102 [62 L.Ed.2d 787, 100 S.Ct. 1067]; see Lummus Company v. Commonwealth *1336Oil Refining Company (2d Cir. 1961) 297 F.2d 80, 89, cert. den., 368 U.S. 986 [7 L.Ed.2d 524, 82 S.Ct. 601].)
The White opinion includes the following quote from United States v. Mespoulede (2d Cir. 1979) 597 F.2d 329, 336-337: “ ‘We tolerate inconsistencies in unified jury verdicts in criminal cases, not because of any singular virtue we attribute to inconsistency, but rather out of deference to the nature of the jury and the role it plays in our jurisprudence. There is no question but that a jury in a criminal trial has the power to render a verdict of acquittal that is wholly at odds with the law and the facts. As we pointed out in United States v. Maybury, 21A F.2d 899, 902 (2d Cir. 1960), this notion has its roots in the fact that the jury was originally conceived of as “inscrutable.” Although we no longer believe that a jury’s pronouncements must be accepted as unquestioningly as the results of an ordeal by cold water or an oath of compurgation, see T. Plunkett, A Concise History of the Common Law 115-16 (5th ed. 1956), an “arbitral” element of jury decision-making survives. We recognize that the jury is in a sense the conscience of the community and can, for example, render a verdict to mitigate an overly severe punishment. United States v. Maybury, supra, 21A F.2d at 902. Similarly, in compromising in order to reach a unanimous verdict, a jury is often fulfilling its role as a cross-section of the community that it is supposed to represent. Occasional anomalies are the price of unanimity. Id., at 903. []]] Internal inconsistency, then, is not an end in itself, and it would be irrational to expand gratuitously the judicial tolerance of inconsistent verdicts to permit different juries in successive trials to reach contradictory results. Allowing a second jury to reconsider the very issue upon which the defendant has prevailed serves no valuable function. To the contrary, it implicates concerns about the injustice of exposing a defendant *1338to repeated risks of conviction for the same conduct, and to the ordeal of multiple trials, that lie at the heart of the double jeopardy clause.’ (Fn. omitted.)” (People v. White, supra, 185 Cal.App.3d at pp. 828-829.)
Penal Code section 954, which applies to this situation, provides that “An accusatory pleading may charge two or more different offenses connected together in their commission, *1339or different statements of the same offense or two or more different offenses of the same class of crimes or offenses, under separate counts .... An acquittal of one or more counts shall not be deemed an acquittal of any other count.” The court concluded that pursuant to this language, a verdict is valid “ ‘even though the jury’s action in returning it was, in a legal sense, inconsistent with its action in returning another verdict of acquittal or guilt of a different offense.’ ” (Lopez, supra, 131 Cal.App.3d at p. 570, quoting Witkin, Cal. Criminal Procedure, § 549, p. 560.)
There is, of course, nothing illogical about finding appellant guilty of murder as an aider and abettor if, as must be presumed to have been the case, the jury was not convinced he used a gun during the crime.
Nunez theorized that if he participated only as an aider and abettor he must have shared the principal’s intent—which a prior jury determined was something less than malice (since Nunez’s coconspirator was convicted of manslaughter, not murder). Similarly, since another jury had concluded that Nunez’s other coconspirator had not committed the killing for financial gain, Nunez argued that he also could not have committed the crime for that purpose.