Opinion

In the published portion of this opinion, we consider several instances of claimed prosecutorial misconduct and examine the scope of vicarious criminal liability of conspirators for the crimes of their confederates in the wake of recent decisions concerning aiding and abetting.

I

George Grant Garewal was convicted by jury of attempted murder, assault with a deadly weapon, mayhem, two counts of burglary, and conspiracy to commit burglary (Pen. Code, §§ 664/187, 245, subd. (a), 203, 459, and 182, subd. 1/459). He was acquitted of robbery, and special enhancements alleging he was armed with a firearm and inflicted great bodily injury were found not to be true (Pen. Code, §§211, 12022, subd. (a), and 12022.7 and 1203.075). The Imperial County District Attorney’s office originally charged Ernest Herrera and Garewal’s then girlfriend, Caroline Chrisman, as codefendants; but Chrisman was given immunity in exchange for her testimony against the other two.

Extensive media coverage and intense public outrage in the relatively small community where the crimes occurred caused venue to be changed to Orange County. The cases against Herrera and Garewal were severed when the latter, as he later admitted, feigned incompetence to proceed (Pen. Code, § 1368). (The irony of the effect of the delay occasioned by Garewal’s charade will not be lost on a reader ¡patient enough to endure the first half of this opinion.) Herrera was tried first; and his conviction was affirmed, as modified, by this court in an unpublished opinion (People v. Herrera (Sept. 26, 1983) 4 Crim. No. 13655).

II

The facts are ugly. On the afternoon of February 14, 1981, Sarah Nicholson and her infant child returned to their home in El Centro, where Garewal, Herrera, and Chrisman were in the process of committing their second residential burglary of the day. When Nicholson drove into her driveway, she blocked the burglars’ vehicle.

At Garewal’s trial, Chrisman testified Nicholson first demanded identification from the three and then sat in her front seat and began to record the license number of their car. Observing this, Herrera commenced a brutal attack, repeatedly striking Nicholson in the face and shattering her eyeglasses. Garewal and Herrera then pulled her roughly from the car; and Herrera, sitting on her supine body, delivered seven or eight more blows with fists, while Garewal went through her purse. Garewal struck her six or seven times in the head as well.

Herrera next obtained a brick and struck Nicholson eight to ten times in the face while Garewal held her hands. The men then started to walk away, but conversed briefly and returned to the motionless victim. Garewal stood on her neck while Herrera viciously kicked her several times in the head. The burglars then departed, leaving Nicholson for dead and the child unattended.

When Herrera entered the car, he stated, “The bitch should be dead, now.” Herrera showed Chrisman a piece of paper and a broken pen. He said, “Sorry Caroline but we had to do that. She can I.D. us.” Later, Garewal told Chrisman only her presence kept him from strangling Nicholson.

The victim and her unharmed child were discovered a short while later. Nicholson’s injuries were not fatal, but they were devastating. She suffered numerous broken fingers and bones and six major skull lacerations and fractures, plus a cerebral concussion. A corneal laceration permanently destroyed virtually all vision in one eye. She walks and talks with great difficulty, has significantly reduced memory recall—including a total inability to remember the attack upon her—and little capacity for cognitive thinking. Routine household chores are almost impossible for her, and she is unable to perform her former job.

Ill

After Garewal was found competent to stand trial, the Orange County Public Defender was substituted for court-appointed defense counsel from Imperial County. In a conversation with the latter, the deputy district attorney volunteered to transport the defense files to Orange County and, in order to assure complete discovery, said he would check them to determine whether any police reports were missing. The prosecutor later explained he was motivated by an office directive to reduce copying costs. Oddly, defense counsel agreed to this procedure without cavil.

Reviewing the files, the deputy district attorney began to read a transcribed conversation which he soon realized was a defense investigator’s incriminating interview of Garewal. He immediately stopped and reported his action to the district attorney. When he turned the files over to Garewal’s new attorney in Orange County the next day, he also informed him of the inadvertent transgression.

Understandably shocked, Garewal’s lawyer sought dismissal for prosecutorial misconduct, arguing the attorney-client relationship had been irreparably violated and it was impossible to be sure the defendant’s statements would not be used to his disadvantage. The trial court conducted a full evidentiary hearing and denied the motion to dismiss, although it did exclude and sanitize the evidence to the fullest extent possible. First, a motion to recuse the Imperial County District Attorney was granted, and the Orange County District Attorney’s office assumed responsibility for the prosecution. Second, the court scrupulously insulated the new prosecutors from even learning of the partial perusal of the Garewal interview, much less the content.

There has been no suggestion that these measures were ineffective, but Garewal nevertheless maintains Barber v. Municipal Court (1979) 24 Cal.3d 742 [ 157 Cal.Rptr. 658 , 598 P.2d 818 ] and United States v. Levy (3d Cir. 1978) 577 F.2d 200 compel dismissal. We disagree. Our Supreme Court found the right to counsel was violated in Barber when an undercover government agent posing as a codefendant infiltrated confidential meetings between misdemeanor defendants and their attorney. Under those circumstances, and specifically because of the chilling effect on the attorney-client relationship, the court found the remedy of exclusion inadequate: “The intrusion, through trickery, of the law enforcement agent in the confidential attorney-client conferences of [defendants] cannot be condoned.” (Barber v. Municipal Court, supra, 24 Cal.3d at pp. 759-760.) Levy reached a similar result, rejecting the notion prejudice need be shown when a police informant posed as a codefendant in attorney-client conferences.

Barber and Levy are inapposite here, however, since both involved intentional intrusions of attorney-client confidences in situations where prejudice to the defendant’s Sixth Amendment rights could not be reasonably measured. (As to the assessment of prejudice in other Sixth Amendment cases, see, e.g., People v. Courts (1985) 37 Cal.3d 784, 796 [ 210 Cal.Rptr. 193 , 693 P.2d 778 ], People v. Joseph (1983) 34 Cal.3d 936, 946 [ 196 Cal.Rptr. 339 , 671 P.2d 843 ], and People v. Smith (1985) 38 Cal.3d 945, 969 [ 216 Cal.Rptr. 98 , 702 P.2d 180 ] (dis. opn. of Bird, C. J.).) When a reviewing court is satisfied that no prejudice could have occurred, suppression is generally found to be an adequate remedy, even where the violation of the defendant’s Sixth Amendment rights was deliberate. (People v. Towler (1982) 31 Cal.3d 105 [ 181 Cal.Rptr. 391 , 641 P.2d 1253 ] [search of defendant’s cell revealed defense strategy and materials]; People v. Glover (1985) 169 Cal.App.3d 689 [ 215 Cal.Rptr. 456 ] [same]; People v. Fulton (1984) 155 Cal.App.3d 91 [ 201 Cal.Rptr. 879 ] [defendant’s legal strategy probed by wired informant during ongoing grand jury proceedings].) As Glover explains, “Given that the evidence seized during the search was never introduced, and thus had no direct effect on the jury’s verdict, in effect, appellant is asking us to reverse his conviction as a sanction for prosecutorial misconduct, rather than to cure any actual prejudice appellant suffered at trial.” (People v. Glover, supra, 169 Cal.App.3d at p. 696 .)

The court added, “While the case at bar is perhaps distinguishable [as is this case] from both Towler and Fulton since in both cases the defendants failed to seek dismissal of the charges upon discovery of the prosecutorial misconduct, and in this case [as here] appellant did seek dismissal, significantly, in both Towler and Fulton , the focus was on what, if any, prejudice resulted from the improper prosecutorial activity. In both cases, where defendant suffered no harm, he was not entitled to a reversal. In Barber , the prejudice suffered by defendants could not be calculated . . . .” (Id., at p. 699.) The record here amply supports the conclusion that the prosecutor’s inadvertent intrusion was not prejudicial to Garewal; consequently, the sanction of dismissal would have been unwarranted.

Moreover, outgoing defense counsel was at least as culpable as the prosecutor in bringing about this bizarre incident. Although we are bewildered by his conduct in voluntarily relinquishing defense files containing confidential incriminating information, knowing the prosecutor would peruse them for missing police reports before they were delivered to Garewal’s new attorney, in view of our holding there is no need to respond to the Attorney General’s suggestion that his action waived any claim of confidentiality—or face the obvious ineffective assistance of counsel problem that ruling would engender.

IV

Before Garewal’s trial began, the defense had cause to renew the motion to dismiss for prosecutorial misconduct. The replacement prosecutor informed new defense counsel of two astonishing developments in the case: First, Chrisman had repudiated some of her former statements and admitted she perjured herself in order to protect Garewal at the defendants’ preliminary hearing, in a deposition, and again at Herrera’s trial.

In those proceedings, she stated Garewal never struck Nicholson, claiming he was wearing a bandage or splint on his hand. She also swore Herrera, not Garewal, placed his foot on the victim’s neck before kicking her. Although she did say Garewal tried to grab Nicholson’s hands at one point, she also asserted he failed. She insisted Garewal never joined in the attack and instead yelled for Herrera to stop.

The prosecutor’s second revelation centered on the timing and circumstances of Chrisman’s “change of heart.” She disclosed her new story to the Imperial County Deputy District Attorney either the same evening or perhaps several days after they engaged in sexual relations at a local motel where the prosecutor and his witnesses from Imperial County were housed during Garewal’s competency trial.

Again the trial court conducted a full evidentiary hearing. Chrisman and the deputy district attorney admitted a single act of intercourse after an evening of dining, dancing, and some ingestion of alcohol on her part. Chrisman denied the sex act with the prosecutor influenced her decision to confess her former perjury and claimed she first told him of Garewal’s true participation in the attack earlier, as they listened to music in a bar after dinner. She lied previously because she feared losing custody of her child and was trying to protect Garewal, whom she considered her common law husband. But since her release from jail, she had given up drugs and thievery and reformed her life. She swore the latest version was the truth.

According to the prosecutor’s recollection, Chrisman first told him of her earlier lies on the drive home to Imperial County a few days after the competency proceedings and the incident at the motel. He denied his personal involvement with Chrisman was motivated by his role as Garewal’s prosecutor and stated he had already decided to ask the district attorney to transfer the case to another deputy because it had become a burden to him. The investigating officer testified Chrisman had also offered to have sex with him on an earlier occasion but he declined.

After careful consideration of the evidence, the trial judge made specific findings: (1) There was no deliberate effort by the prosecution to undermine Garewal’s defense: “[Wjhat we are dealing with here is a situation of one man’s misconduct and that is all. . . . [Hjis actions are a combination of poor judgment, possibly unethical conduct, possibly some stupidity; but I don’t see any indication in this record of an intent to deprive Mr. Garewal of his rights.” (2) There would be little damage to Garewal’s case: “[I]f anything, the whole thing has worked to Mr. Garewal’s advantage. He should be able to thoroughly impeach Miss Chrisman as a witness in light of everything that has gone on.” (3) There was no connection between Chrisman’s change of testimony and the sexual episode, and no emotional attachment between the prosecutor and the witness had developed. Chris-man’s explanation was credible; i.e., she had changed her life and no longer desired to protect Garewal. Finally, the judge noted the deputy district attorney and his office had already been removed from the case, and the public interest in prosecuting an offense as serious as that before the court weighed heavily against dismissal in the absence of convincing evidence that the defendant’s constitutional rights had been impaired.

It is suggested for the first time on appeal that the prosecutor may have caused Chrisman to recant based on something he told her of the incriminating statement he read in the defense file. There is not a scintilla of evidentiary support in the record for such a notion, however; all the evidence indicates the change of story preceded the incident with the files. Indeed, no questions were even asked at the hearing on the renewed motion to dismiss concerning this contention. Consequently, we decline to infer the existence of more ghosts than are already known to haunt this particular house.

Nevertheless, Garewal argues he is entitled to a reversal with directions to dismiss based on the prosecutor’s misconduct or, in the alternative, retrial with suppression of Chrisman’s testimony. Since the trial court’s findings are based on substantial evidence, we are unable to agree. If Chrisman’s testimony had changed after being seduced for that purpose, severe sanctions would have been indicated (see, e.g., People v. Warren (1984) 161 Cal.App.3d 961 [ 207 Cal.Rptr. 912 ]); and it is clear the trial court was prepared to apply them. But according to the testimony before the court, Chrisman was not seduced any more than the amorous deputy district attorney was, and perhaps less. The trial court had ample reason to find her decision to risk a perjury prosecution was not a product of the tryst with the prosecutor.

The deputy district attorney’s behavior certainly would have justified his removal from the case, but his whole office had already been dispatched before this sordid little scenario came to light. In view of all that had occurred, the uniquely appropriate means of evaluating Chrisman’s testimony was the one selected, a jury trial. The effect of love and sex on credibility is well within the range of common experience and, unlike hypnotism for example, as subject to jury evaluation as any other common form of human behavior. (Cf. People v. Guerra (1984) 37 Cal.3d 385 [ 208 Cal.Rptr. 162 , 690 P.2d 635 ].)

Finally, hindsight validates the trial judge’s prediction: Garewal’s counsel used Chrisman’s former testimony for a telling impeachment of the witness. Also, as will be discussed below, the verdicts indicate the jury may well have rejected the new gloss on her testimony (if it found any reason to consider it at all in view of the instructions given by the court).

V

VII

Garewal attacks several of the court’s instructions to the jury. He first claims the evidence demonstrated the conspiracy to commit burglary ended before the assault on Nicholson began and thus could not support an instruction that he might be held vicariously liable for Herrera’s actions as part of a conspiracy to commit burglary. The contention is meritless. Crimes such as robbery and burglary are not considered legally complete for purposes of the felony-murder rule, for example, until the criminals have reached a place of temporary safety and achieved unchallenged possession of the loot. (People v. Salas (1972) 7 Cal.3d 812, 821-823 [ 103 Cal.Rptr. 431 , 500 P.2d 7 , 58 A.L.R.3d 832 ], cert, den., sub nom., Salas v. California (1973) 410 U.S. 939 [ 35 L.Ed.2d 605 , 93 S.Ct. 1401 ]; People v. Scott (1985) 170 Cal.App.3d 267, 271 [ 215 Cal.Rptr. 618 ].)

Similarly, a conspiracy to commit a particular crime concludes no earlier than the legal completion of the intended offense itself. The rule is a hoary one: “[E]ach conspirator is bound by the acts of a confederate in furthering the common design of the conspiracy by escaping or resisting arrest, even though such acts may have been ‘dictated by the exigencies of the moment.’ ” (People v. Smith (1966) 63 Cal.2d 779, 794 [ 48 Cal.Rptr. 382 , 409 P.2d 222 ], cert, den., sub nom., Smith v. California (1967) 388 U.S. 913 [ 18 L.Ed.2d 1353 , 87 S.Ct. 2119 ]; citing cases, e.g., People v. Kauffman (1907) 152 Cal. 331 [ 92 P. 861 ].)

VIII

The next assignment of instructional error is far more troublesome, however. Garewal raises serious questions concerning the scope of vicarious liability of aiders and abettors and conspirators for the incidental but unplanned crimes of confederates in pursuit of the common goal.

As a preface to our discussion of the instructions relating to vicarious liability, we note the evaluation of any error concerning them presupposes rejection of the changes in Chrisman’s testimony at Garewal’s trial. By the account she gave there, he was a direct perpetrator in the attack; and conspiracy and aiding and abetting instructions were surplusage in light of that version with respect to the substantive counts. (People v. Hayes (1985) 169 Cal.App.3d 898, 911 [ 215 Cal.Rptr. 595 ]; People v. Talamantez (1985) 169 Cal.App.3d 443, 462-463 [ 215 Cal.Rptr. 542 ].) Unfortunately, it is concededly impossible to ascertain from the verdicts how the jury reacted to the revisions of Chrisman’s recitation; and the prosecutor strongly argued Garewal’s vicarious responsibility on conspiracy and aiding and abetting theories.

The scope of an aider and abettor’s vicarious responsibility for the actions of the perpetrator has undergone close examination in the recent legal history of this state. In Carlos v. Superior Court (1983) 35 Cal.3d 131 [ 197 Cal.Rptr. 79 , 672 P.2d 862 ], the Supreme Court interpreted the applicable statutes to permit imposition of the death penalty or life without possibility of parole only if the aider and abettor shared the perpetrator’s intent to kill. As the court stated in a subsequent case, “Defendant did not shoot the victim and was not present in the store at the time of the shooting. The fact that he agreed to aid a robbery knowing his companion was armed, is insufficient to demonstrate that defendant himself intended to aid a killing. (Carlos v. Superior Court, supra, 35 Cal.3d 131, 151 ; see Enmund v. Florida (1982) 458 U.S. 782 [ 73 L.Ed.2d 1140 , 102 S.Ct. 3368 ].)” (People v. Garcia (1984) 36 Cal.3d 539, 557 [ 205 Cal.Rptr. 265 , 684 P.2d 826 ], cert, den., sub nom., California v. Garcia (1985) — U.S. — [ 84 L.Ed.2d 366 , 105 S.Ct. 1229 ].)

Next the court held the standard aiding and abetting instructions, CALJIC Nos. 3.00 and 3.01, given here at the request of both sides, incorrectly state the law because they fail to limit liability to those who share the intent of the perpetrator. (People v. Beeman (1984) 35 Cal.3d 547 [ 199 Cal.Rptr. 60 , 674 P.2d 1318 ]; see also People v. Caldwell (1984) 36 Cal.3d 210 [ 203 Cal.Rptr. 433 , 681 P.2d 274 ].) The Beeman rule is retroactive. (People v. Olmedo (1985) 167 Cal.App.3d 1085, 1091-1092 [ 213 Cal.Rptr. 742 ]; People v. Henderson (1985) 163 Cal.App.3d 1001, 1009 [ 209 Cal.Rptr. 883 ]; People v. Minichilli (1984) 161 Cal.App.3d 660, 669 [ 207 Cal.Rptr. 766 ]; People v. Gilman (1984) 156 Cal.App.3d 760, 764-765 [ 203 Cal.Rptr. 6 ].)

Initially, we must consider whether Garewal should even be permitted to question the aiding and abetting instructions. Failure to object to the instructions condemned in Beeman and Caldwell has been held not to constitute invited error which forecloses the issue on appeal (People v. Henderson, supra, 163 Cal.App.3d at p. 1009 ); but defense counsel here did not simply acquiesce in them. He joined in the request—and long after they were first questioned in appellate opinions. (See, e.g., People v. Yarber (1979) 90 Cal.App.3d 895 [ 153 Cal.Rptr. 875 ].)

To our knowledge no court has considered whether these instructions— when given at the defendant’s request—constitute invited error. Ordinarily, of course, we would hold a specific request for an instruction waives any error in giving it. In this rare instance we are not so inclined, however. There are several reasons:

First, defense counsel would have been remiss had he not, as a minimum, sought the standard CALJIC aiding and abetting instructions. At least they warned that an aider and abettor present at the scene of a crime must act with knowledge of the perpetrator’s intentions if he is to share in the responsibility.

Second, it is difficult to lay blame on defense counsel when the prosecution and perhaps most trial and appellate courts believed the standard instructions were sufficient. For example, in Beeman itself the Attorney General argued “that the standard instruction properly reflects California law, which requires no more than that the aider and abettor have knowledge of the perpetrator’s criminal purpose and do a voluntary act which in fact aids the perpetrator.” (People v. Beeman, supra, 35 Cal.3d at p. 555; see cases cited at pp. 555-557.)

Moreover, six months before this case was tried and as late as 18 months after, our own appellate district was cynically rejecting the Yarber court’s criticism of CALJIC No. 3.01: “This court follows the general rule following a long line of cases finding CALJIC No. 3.01 sufficient. ‘As is said in People v. Ott (1978) 84 Cal.App.3d 118 , at page 130 . . ., “aiding in the commission of the crime with knowledge of the wrongful purpose of the perpetrator eo ipso establishes the criminal intent ... ‘the criminal intent of the aider and abettor is presumed from his actions with knowledge of the actor’s wrongful purpose’ (italics added)” [citation]. We might add to these truths the observation that since our system of justice entrusts to the jurors the vast power of drawing any inference that can reasonably be drawn from substantial evidence and we foreover [sz'c] bind ourselves to those inferences properly drawn, it would be demeaning of jurors’ intelligence to instruct them, in addition, on the clear implication of intent that flows logically and necessarily from the words of the CALJIC No. 3.01 standard instruction. The standard CALJIC No. 3.01 is all that is required.’ (People v. Flores (1982) 128 Cal.App.3d 512, 525 [ 180 Cal.Rptr. 368 ].)” (In re Martin (1983) 150 Cal.App.3d 148, 157 [ 197 Cal.Rptr. 655 ].) We can hardly penalize Garewal for his counsel’s failure to read the leaves in this particularly murky cup of tea or for his submission to the wisdom of our own appellate district in not demeaning the intelligence of jurors.

Moreover, even if he had sought and obtained the Yarber modifications, there would be no difference. The Yarber instructions were themselves criticized as insufficient in Beeman ; and authority from this district now holds it is reversible error to give the standard aiding and abetting even amended as suggested in Yarber. (People v. Henderson, supra, 163 Cal.App.3d at p. 1008 .)

Finally, Penal Code section 1259 provides an “appellate court may . . . review any instruction given, refused or modified, even though no objection was made thereto in the lower court, if the substantial rights of the defendant were affected thereby.” (People v. Satchell (1971) 6 Cal.3d 28, 33, fn. 10 [ 98 Cal.Rptr. 33 , 489 P.2d 1361 , 50 A.L.R.3d 383 ]; People v. Henderson, supra, 163 Cal.App.3d at p. 1009 .) In view of the conspiracy instruction we consider next, which greatly exacerbated the Beeman error, this is such a case; for, as another court put it, “The jury here was not instructed it must find both [the defendant] and the stabber harbored a specific intent to kill in order to convict [the defendant] of attempted murder. The jury in fact was instructed to the contrary . . . .” (People v. Acero (1984) 161 Cal.App.3d 217, 224 [ 208 Cal.Rptr. 565 ].)

CALJIC No. 6.11 correctly defines the vicarious legal responsibility of coconspirators. Here, however, it was modified by the court at the prosecution’s urging. The instruction, with the amendment emphasized, reads, “Each member of a criminal conspiracy is liable for each act and bound by each declaration of every other member of the conspiracy if said act or said declaration is in furtherance of the object of the conspiracy, [t] The act of one conspirator pursuant to or in furtherance of the common design of the conspiracy is the act of all conspirators. Every conspirator is legally responsible for an act of a coconspirator that follows as one of the probable and natural consequences of the object of the conspiracy even though it was not intended as a part of the original plan, or was even actually forbidden as part of the original agreement and even though he was not present at the time of the commission of such act.”

Thus, under this instruction, the jury had to resolve but one question of intent concerning the substantive offenses: Did Garewal conspire to commit burglary? If so, he was automatically and strictly liable for all that followed, even acts contrary to the agreement with his cohorts.

We have encountered no other post-Beeman case where, as here, the defendant’s vicarious responsibility for certain crimes was also potentially predicated on a theory of conspiracy to commit a different crime, i.e., burglary. (Cf. People v. Gilman, supra, 156 Cal.App.3d 760 [Beeman error in instruction concerning mayhem held harmless because defendant was convicted of conspiracy to commit mayhem under other appropriate instructions].) Without the wording added by the court, CALJIC No. 6.11 is quite consistent with Beeman , which recognizes an aider and abettor’s “liability . . . extends also to the natural and reasonable consequences of the acts he knowingly and intentionally aids and encourages. (People v. Beltran [1949] 94 Cal.App.2d 197, 207 [ 210 P.2d 238 ].)” (People v. Beeman, supra, 35 Cal.3d 547, 560 .) But the additional language makes the instruction logically inconsistent, for how can one anticipate as a “probable and natural consequence[] of the object of the conspiracy” an act which was “actually forbidden as part of the agreement”? Worse, it extends a relatively mild form of vicarious liability, one which is at least limited by the reasonable contemplation of the defendant, although perhaps not by his intent, to acts which are specifically not contemplated, much less intended.

We are aware a good argument could have been made—at least before Beeman—that the court’s amendment correctly stated the law. Witkin said so: “Liability extends to acts unintended or even actually forbidden, if they are, nevertheless, in furtherance of the common purpose.” (1 Witkin, Cal. Crimes (1963) § 121, p. 114, citing cases.) Several subsequent opinions expressly agreed (People v. Manson (1976) 61 Cal.App.3d 102, 210 [ 132 Cal.Rptr. 265 ], cert, den., sub nom., Manson v. California (1977) 430 U.S. 986 [ 52 L.Ed.2d 382 , 97 S.Ct. 1686 ], and People v. Beaumaster (1971) 17 Cal.App.3d 996, 1003 [ 95 Cal.Rptr. 360 ]); but the facts of those cases supported no such sweeping concept of criminal responsibility. There was not the slightest hint in either of an agreement barring an act subsequently committed in furtherance of the actual object of the conspiracy.

Both Manson and Beaumaster rely upon People v. Smith, supra, 63 Cal.2d 779, 794 ; but Smith neither states nor stands for the proposition that coconspirators are strictly liable for each other’s acts. Indeed, the discussion of conspiracy in Smith is dicta, for it was reversed on other grounds as to the passive defendants and affirmed without need to resort to conspiracy theory as to the perpetrator. (Id., at p. 792, fn. 7.) In any event, Smith is a felony-murder case, albeit a somewhat unusual one, and consequently poor authority as to the scope of the legal liability of coconspirators. Intent and foreseeability are of no moment in felony-murder prosecutions.

Probably the best known opinion in our national jurisprudence on the subject of the liability of conspirators is Pinkerton v. United States (1946) 328 U.S. 640 [ 90 L.Ed. 1489 , 66 S.Ct. 1180 ]. One commentator explains the holding as follows: “It should be noted that Justice Douglas was careful in Pinkerton not to make the conspiracy-complicity rule a strict liability doctrine. Thus, it is a defense for a conspirator to demonstrate that the ‘substantive offense committed by one of the conspirators was not in fact done in furtherance of the conspiracy, did not fall within the scope of the unlawful project, or was merely a part of the ramifications of the plan which could not be reasonably foreseen as a necessary or natural consequence of the unlawful agreement.’ (Id., at pp. 647-648.)” (Marcus, Prosecution and Defense of Criminal Conspiracy Cases (1983) § 7.03[2], p. 7-12.)

Even this limitation on liability has been criticized as insufficient, however: “If the Pinkerton rule were adhered to, each prostitute or runner in a large commercialized vice ring could be held liable for an untold number of acts of prostitution by persons unknown to them and not directly aided by them. Each retailer in an extensive narcotics ring could be held accountable as an accomplice to every sale of narcotics made by every other retailer in that vast conspiracy. Such liability might be justified for those who are at the top directing and controlling the entire operation, but it is clearly inappropriate to visit the same results upon the lesser participants in the conspiracy.” (LaFave & Scott, Criminal Law (1972) § 65, p. 514; see People v. Luciano (1938) 277 N.Y. 348 [ 14 N.E.2d 433 ], cert, den., sub nom., Luciano v. New York (1938) 305 U.S. 620 [ 83 L.Ed. 396 , 59 S.Ct. 81 ].)

Another commentator expressed a similar thought: “The difficulty lies not in the conspiracy-complicity rule itself, but in the tendency of courts to regard a conspiracy as an ongoing business relationship of indefinite scope and duration, and to consider the conspirators, as one dissenting opinion put it, as ‘general partners in crime. ’ [Rutledge, J., dis. in Pinkerton v. United States, supra, 328 U.S. at p. 651 .] For example, the defendant in Anderson v. Superior Court [(1947) 78 Cal.App.2d 22 ( 177 P.2d 315 )] referred several pregnant women to an abortionist and received a portion of his fees. For this the court held her to have entered into a conspiracy with him to commit abortions generally, and to be liable for subsequent abortions in which she played no part.” (Johnson, The Unnecessary Crime of Conspiracy (1973) 61 Cal.L.Rev. 1137, 1147.)

Fortunately, we need not consider the continuing viability of Anderson . There, at least, the object of the conspiracy was the commission of abortions, and “[n]o one would question that all the persons who plot together to commit a crime are guilty of the crime if one or more of them commits it. [Not so; see, e.g., Commonwealth v. Perry (Mass. 1970) 256 N.E.2d 745, 747 .] Some authorities limit the accomplice’s liability to those crimes of the principal which he intended to assist or encourage. [Citing Pinkerton and Anderson.] Many other authorities, however, have indulged in the legal fiction that one intends the natural and probably [i/c] consequences of his acts, and thus have held the accomplice for the crimes of the principal which he should have foreseen but perhaps did not.” (Id., at pp. 1146-1147.) California imposes liability under all these theories, including the “legal fiction.” For example, in Beeman itself the court specifically explains the responsibility of an aider and abettor extends to “reasonable and probable consequences. ”

Also, the Beeman opinion mixes citations to conspiracy and aiding and abetting cases without distinguishing between them. In concluding that “the weight of authority and sound law require proof that an aider and abettor act with knowledge of the criminal purpose of the perpetrator and with an intent or purpose either of committing, or of encouraging or facilitating commission of, the offense,” the court cites People v. Terry (1970) 2 Cal.3d 362, 402 [ 85 Cal.Rptr. 409 , 466 P.2d 961 ], cert, den., sub nom., Terry v. California (1972) 406 U.S. 912 [ 32 L.Ed.2d 112 , 92 S.Ct. 1619 ]. (People v. Beeman, supra, 35 Cal.3d at p. 560.) Terry is a felony-murder case; and on the page referenced above, the Supreme Court finds the following to be a correct statement of the law: “[W]here a conspirator [or one who is acting jointly in the commission of a crime with another] commits an act which is neither in furtherance of the conspiracy or the intended crime nor the natural and probable consequence of an attempt to attain that object, he alone is responsible for and is bound by that act, and no responsibility therefor attaches to any of his confederates.” (People v. Terry, supra, 2 Cal.3d at p. 402, fn. 18 .) Although the quotation is subject to a possible interpretation that a coconspirator is responsible for any crime committed in furtherance of the conspiracy, whether or not “natural and probable,” it is clear that Beeman does not so interpret it. As the opinion remarks in a related context, “we believe this is misconstruction of Terry . ” (People v. Beeman, supra, 35 Cal.3d at p. 557.)

It is, of course, possible to devise a model which would provide a wider scope of criminal liability for conspirators than aiders and abettors. But Beeman does not differentiate between them. Also, the great majority of cases involving aiding and abetting would also lend themselves to conspiracy instructions. (See Perkins, The Act of One Conspirator (1974) 26 Hastings L.J. 337 , 340.) We find the notion that Beeman might be so easily avoided highly dubious. To the contrary, we conclude the clear thrust of Beeman is to contain the reach of vicarious criminal responsibility of conspirators to the natural and reasonable consequences of the conspiracy.

Finally, it remains to determine whether the combined instructional error in this case requires reversal. The convictions for the two burglaries and conspiracy to commit burglary are not affected because Garewal’s exposure for those offenses was always as a direct perpetrator, not as an aider and abettor. (People v. Hayes, supra, 169 Cal.App.3d 898, 911 .) But the same is not true as to the counts relating to the attack on Nicholson.

Although the question of which standard to be applied with respect to Beeman error is still an open one, there is no reason to join the current debate on the subject. We find the error here, as in Beeman itself, would not survive even the most conservative test, that of People v. Watson (1956) 46 Cal.2d 818 [ 299 P.2d 243 ]. (See People v. Tovar (1984) 161 Cal.App.3d 137, 139, 142 [ 207 Cal.Rptr. 255 ].)

As the parties agree, we may only speculate on the jury’s ultimate conclusions regarding Chrisman’s testimony at Garewal’s trial. The two find ings that he did not personally inflict great bodily injury on the victim do tend to support the view that her new version of the attack was rejected; but this is not necessarily so because the court instructed the jury to “determine whether or not the defendant with the specific intent to inflict [great bodily injury], did personally inflict great bodily injury. . . . [S] The term ‘personally ’ as used in this instruction, means that the defendant must have directly performed the act or acts which caused the physical injury to the victim. It does not include one who merely aids, abets, or directs another person to do so.” (CALJIC No. 17.20 (1979 revision), as amended by the court in the italicized portions.) Thus, the combination of the conviction for mayhem and rejection of the great bodily injury allegations can be explained by the stringent caveat requiring personal infliction with respect to the latter, though it seems unlikely the jury could have believed Garewal personally struck the face of the victim yet entertained a reasonable doubt as to the great bodily injury allegations.

In sum, it is exceedingly possible, even probable, that Garewal was convicted per the court’s instructions and the prosecutor’s argument as an aider and abettor or coconspirator in the crime of burglary without any further consideration of his intent or his reasonable expectations. Our decision is thus compelled: “[W]hen the prosecution (or the court) presents the case to the jury on alternate theories, some of which are legally correct and others legally incorrect and the reviewing court cannot determine from the record on which theory the ensuing general verdict of guilt rested, the conviction cannot stand. [Citations.]” (People v. Birreuta (1984) 162 Cal.App.3d 454, 462 [ 208 Cal.Rptr. 635 ].) As to the attempted murder, assault with a deadly weapon, and mayhem counts, we are unable to divine which theory the jury adopted; they must be retried under appropriate instructions. (People v. Henderson, supra, 163 Cal.App.3d at p. 1011 .)

The burglary and conspiracy convictions are affirmed, but the judgments must be vacated because reversal of the base term offense may invalidate the entire sentencing scheme. The remaining convictions are reversed, and the case is remanded accordingly.

Trotter, P. J., and Wallin, J., concurred.

Respondent’s petition for review by the Supreme Court was denied February 14, 1986. Mosk, J., Lucas, J., and Panelli, J., were of the opinion that the petition should be granted.

See footnote, ante, page 285.