HANDLER, J., concurring in part and dissenting in part.

Defendant, Jacinto Koger Hightower, whose conviction for murder has previously been affirmed by this Court, State v. Hightower, 120 N.J. 378, 577 A.2d 99 (1990) ( Hightower I), now appeals from the sentence of death imposed by a jury at his resentencing trial. The facts of the crime are reported in some detail in this Court's opinion on defendant's previous appeal, and in the majority's opinion.

Hightower's jury was on the verge of returning a verdict that would have resulted in a sentence of life imprisonment when the trial court intervened and removed the one juror poised to resist a death verdict. The juror had engaged in misconduct — he communicated extra-record information concerning the victim to the deliberating jury. I concur with the majority's conclusion that the trial court abused its discretion by removing that juror and reconstituting the jury with a substitute juror because the deliberations had proceeded too far. Because the information prejudiced only defendant, the Court's conclusion that the trial court should have declared a mistrial, and that as a result of the failure to declare a mistrial, defendant must, for a third time, be exposed to the death sentence is draconian. Other options were available to the court in dealing with this form of misconduct. The court could have allowed the juror to remain on the jury or direct a verdict of a life sentence. Not having done either, defendant irretrievably lost the opportunity of a final non-unanimous verdict and the resultant life sentence. Fundamental fairness and constitutional precepts dictate that the death sentence be set aside and a life sentence imposed.

I dissent from this case for another reason. The majority concludes that the evidence below was sufficient to support an aggravating factor, c(4)(f), which involves a murder committed to avoid apprehension. To sustain that determination the majority has so stretched the meaning of c(4)(f) that it can be applied arbitrarily to virtually any felony murder.

I

A.

The facts relating to the juror misconduct issue are as follows:

On the second day of jury deliberations, at 2:55 p.m., the trial court announced receipt of the following note from the jury:

Judge, it has come to our attention that one of the jurors has information and knowledge about this case beyond the scope of the evidence given in Court. We ask your direction in regard to this matter.

At the time the note was written, the jury had already spent approximately six and one-half hours deliberating. The court suspended deliberations and then proceeded to explore the nature and scope of the problem through individualized voir dire of the jury.

During the court's questioning of the jury foreman, several significant facts came to light. First, the court learned that Juror Number Seven told the rest of the jury that he had heard that the victim had three children. The foreman further explained the context that provoked Juror Seven's revelation:

There was a fairly intense debate going on at that point in time where [Juror Seven] was taking a different position than some of the other people, and there was a fair amount of frustration from some other people about the position he was taking, and essentially he was — if I can characterize, he was getting backed into a corner, so to speak, or something like that and he was almost lashing out. This is the way that I read it, and saying, look, I know stuff.

The essential facts of this account were confirmed by the other jurors, although there was extensive confusion concerning the details of Juror Number Seven's comment. There was also significant confusion among the jury concerning the source of Juror Seven's information, and even whether Juror Seven divulged his source.

Juror Number Seven told the court that he overheard the information while food shopping earlier that week. He denied discussing the case with anyone, and even stated that he refused to speak about the case when approached by his son. Juror Number Seven explained the reason for his disclosure. He said:

I told them I thought that he should get 30 years imprisonment instead of the death sentence because it seems by his time in prison he could be a productive citizen when he comes out. Now, majority of them jumped me for that. . . . Well, they said, you must have heard something we didn't hear, and I says, all I overheard — I didn't discuss this with no one — was that the lady had three children. . . .

From the foreman, the court and counsel learned the true extent to which Juror Seven was alone in his view of the appropriate penalty. The foreman stated:

Let me give you a little bit fuller context and trust that — I guess I can do that. [Juror Seven] was taking a different position than others in the room. . . . And it's frustrating to me because he has a point of view that deserves to be on the jury, and what I'm — what I'm afraid of is that if — jumping to a conclusion here — that he might be removed or replaced with an alternate or something like that, that his point of view gets lost.

The court also discovered that jury deliberations had reached a very late, if not conclusive stage at the time of the interruption. From the colloquy at the close of the previous day's deliberations it was apparent that the jury had entered the decision-making stages because it had already begun to mark the verdict sheet. During the course of the questioning the foreman explained:

We have conducted a great jury deliberation and it's — and it's frustrating to me that at this — which was a fairly late point in our deliberation that this came out.

Juror Number Two further hinted at the advanced state of deliberations when, in answer to a question about whether the information that the victim had children would affect him, he replied "[a]s far as I'm personally concerned, no. I made up my mind in a sense — whatever." Juror Number Three added to this impression by stating that the information would not affect her "because prior to [Juror Number Seven] saying that, we had already taken a straw vote." Finally, Juror Number Ten made a comment about "when we were voting on the mitigating circumstances. . . ." These comments demonstrate that the jury had advanced very far in its deliberations.

Significantly, counsel for the defense twice requested that the court inquire of the foreman as to what stage in the deliberative process the jury had reached when the information was revealed and once requested that the court poll the jurors individually as to their capacity to start deliberations anew with a substitute juror. The court refused to undertake those inquiries, expressing confidence in the jury's ability to restart its deliberations and to disregard its past conclusions as well as the irrelevant victim-impact evidence. The court thus denied the defense motion for a mistrial, and removed Juror Number Seven from the jury. Juror Number Seven was replaced with an alternate juror, and the eleven old and one new jurors were told to begin deliberations anew.

The reconstituted jury commenced deliberations the following morning at 11:03 a.m. At 3:30 p.m., the jury returned a verdict sentencing Hightower to death. The new deliberations lasted four and one-half hours.

By Order, this Court unsealed the first jury verdict sheet. The markings on the sheet indicated that the jury had found both aggravating factors and had made findings on all of the fourteen mitigating factors. The only section of the sheet left unmarked was that reserved for the indication of the jury's ultimate conclusion on the sentence.

B.

Extreme restraint and caution must be employed whenever a court acts in a fashion that might alter or influence jury deliberations. In State v. Corsaro, 107 N.J. 339 , 526 A.2d 1046 (1987), the Court explained that "[i]n light of the centrality of jury deliberations to our criminal justice system, errors that could upset or alter the sensitive process of jury deliberations, such as improper juror substitution, 'trench directly upon the proper discharge of the judicial function.'" Id. at 347, 526 A.2d 1046 (citing State v. Harper, 128 N.J. Super. 270, 278, 319 A.2d 771 (App.Div.) (citation omitted), certif. denied, 65 N.J. 574 , 325 A.2d 708 (1974)).

The majority correctly concludes that the trial court's decision to remove Juror Number Seven was an abuse of the court's discretion because the reasons for removal were not exclusively personal circumstances as required by Rule 1:8-2(d). Ante at 255-56, 680 A.2d at 656-57 (quoting State v. Trent, 157 N.J. Super. 231, 239, 384 A.2d 888 (App.Div. 1978), rev'd on other grounds, 79 N.J. 251 , 398 A.2d 1271 (1979)). After concluding that the trial court abused its discretion in removing Juror Number Seven, the majority significantly discounts or ignores the crucial factual circumstances of this case as well as the alternative remedies available to the trial court in holding that the only remedy was a mistrial.

Generally, when a juror has committed misconduct but removal and substitution is not a viable option, a trial court may consider either retaining the juror and providing a curative instruction, or declaring a mistrial. The court should apply the least drastic remedy, and the remedy which would least infringe on a defendant's right to a fair trial. As the majority correctly notes, in a capital cause, a court must be extremely cautious in guarding a defendant's fundamental right to fairness. Because of the unique circumstances in the present case, the majority's conclusion that a mistrial would have been the appropriate remedy prior to any further voir dire on the jurors' ability to continue is incorrect and commits an injustice upon defendant that our system cannot tolerate.

A mistrial is warranted if juror misconduct "could have a tendency to influence the jury in arriving at its verdict in a manner inconsistent with the legal proofs and the court's charge." Panko v. Flintkote Co., 7 N.J. 55 , 61, 80 A.2d 302 (1951). The majority holds that " any irregularity [in the deliberative process] is presumptively prejudicial." Ante at 267, 680 A.2d at 663 (emphasis added) (citing Corsaro, supra, 107 N.J. at 346, 526 A.2d 1046 (citation omitted) (noting that errors of juror substitution are cognizable as plain error); State v. Auld, 2 N.J. 426 , 432, 67 A.2d 175 (1949)). The majority also highlights that this standard must be strictly observed in the unique context of a death-penalty case where the stakes are extraordinary. Unfortunately, the majority does so only on an abstract level, ignoring the context and atmosphere that surrounded the deliberations of this jury when the misconduct arose.

In all cases, however, it is necessary to be mindful that a mistrial is an extreme remedy that should be invoked only as a last resort. Not only does a retrial result in a great expenditure of time, money, and judicial resources, State v. Lipsky, 164 N.J. Super. 39 , 43, 395 A.2d 555 (App.Div. 197 8), but, in a capital prosecution when the jury is poised to return a verdict that would result in a life sentence, the remedy itself becomes a life-or-death decision.

C.

A mistrial was neither the necessary nor the appropriate remedy in this case. The jury was not irremediably tainted as a result of its exposure to the collateral extra-record information and Juror Number Seven's conduct did not mandate his removal from the jury. Most significantly, prior to the introduction of the extra-record information, the jury was poised to reach a non-unanimous verdict.

The trial court itself felt very strongly that the jury was sincere in reporting its ability to continue its deliberations. After conducting individualized voir dire of each juror regarding his or her ability to continue, the court concluded that the jury could resume deliberations without considering the improper disclosure of victim impact information. The inference is clear that the jury was not tainted by the extra-record information.

The trial court should have determined, though, that if a curative instruction was sufficient for the jury, it was also sufficient for Juror Number Seven. Juror Number Seven demonstrated that he had, in fact, disregarded the extra-record information. Indeed, the only juror for whom the trial court had direct proof as to whether the information had an effect is Juror Seven. Knowledge that the victim was survived by children, when combined with all the victim-impact inferences engendered by that fact, militates, if at all, toward the imposition of the death penalty. Juror Seven was, however, convinced that Hightower should be sentenced to life. Therefore, it is clear that the information to which Juror Seven was exposed did not manifestly affect his analysis of any of the issues before the jury.

Nor did Juror Seven's behavior amount to misconduct compelling his removal from the panel. This was not a traditional case of juror misconduct. Juror Seven overheard the information, he did not seek it out. The trial court did not find that the juror discussed the fact with others outside of the court. There is no evidence, nor did the court find, that the disclosure was deliberate or premeditated. Rather Juror Seven blurted out the statement as a result of difficult interaction with fellow jury members, in an improper but not ill-motivated attempt to defend his position. The disclosure was impulsive and intemperate, not calculated or designed. The pro-life juror did not reveal facts to sabotage the deliberations, but was actually defending his views by asserting the strength of his conviction despite his knowledge of extra-record, pro-death facts. It was clear that Juror Seven could have prevented the imposition of the death sentence by simply withholding his vote. Instead, he chose to respect an affirmative obligation to deliberate, to attempt to convey his point of view to the other jurors so they could analyze and weigh his perspective. Finally, Juror Number Seven did not demonstrate contempt for the court, but was honest and forthright when confronted with his transgression. Because of the nature of his revelation and the context in which it was uttered, there was little likelihood that Juror Seven was going to repeat his behavior.

State v. Ramseur, 106 N.J. 123 , 524 A.2d 188 (1987), is instructive in exploring the scope and the importance of the right to a life verdict. In Ramseur, the Court recognized that "the Legislature contemplated three possible final verdicts in a capital case": a unanimous verdict for death, a unanimous verdict for life, and a non-unanimous verdict that results in a life sentence. Id. at 301, 524 A.2d 188. The Court held that fundamental fairness requires courts to respect even the potential opportunity for a non-unanimous jury.

After four hours of deliberation, Ramseur's jury advised the trial court that the jury was "unable to reach a unanimous decision. Suggestions please." Ibid. The Court in Ramseur held that the note did not necessarily indicate a deadlocked jury. Rather, in seeking direction, the jury communicated that it could continue. However, the Court was aware of a strong potential for deadlock. The Court suggested that the trial court "should have explored with the jury whether it had deliberated sufficiently and had reached a genuine stalemate, a point at which further deliberations would have been counter-productive." Id. at 304, 524 A.2d 188. The trial court, however, made no attempt to discover whether further deliberations would be helpful. Rather, the trial court immediately gave an Allen charge. Id. at 305, 524 A.2d 188 (citing State v. Czachor, 82 N.J. 392, 402, 413 A.2d 593 (1980) (citing State law grounds and disapproving of traditional Allen charge); Allen v. United States, 164 U.S. 492, 17 S.Ct. 154, 41 L. Ed. 528 (1896) (approving jury charge instructing jury that "it was their duty to decide the case if they could conscientiously do so" and "they should listen, with a disposition to be convinced, to each other's arguments")). The Ramseur Court held that the Allen charge had a potential to coerce the jury to return a unanimous verdict because the effect of telling the jury that it had to agree would undermine any likelihood it would disagree. Ibid. Because the jury may have ultimately returned a verdict for life had the coercive charge not been delivered, the Court reversed the death sentence. The Court held that the defendant could not be subjected to the death penalty in a second trial.

The trial court in this case faced a situation that paralleled that of the trial court in Ramseur: (1) the court was faced with strong evidence of the jury's "unwillingness to bring in an uncoerced unanimous verdict for the death sentence," id. at 313, 524 A.2d 188; (2) the trial court did not inquire into whether the jury had reached a stalemate or whether further deliberations would be helpful; and (3) error by the trial court prevented the court from discovering the effect of further deliberations.

Here, there are strong indications that the jury would not have been unanimous. As a result of the voir dire, the trial court knew that Juror Number Seven was poised to vote for life. Moreover, the court knew that Juror Number Seven was forcefully committed to that position. The court also knew that many, and possibly all of the other jurors had already made up their minds to return a death sentence. Indeed, the court was warned that if Juror Seven was removed from the panel, "his point of view gets lost." The court also knew deliberations were at a late stage and that the jury was experiencing great frustration with the strong-minded Juror Number Seven's commitment to advocate for life.

When a communication from the jury indicates that it may have reached a stalemate, a trial court should ascertain whether further deliberations would be helpful to the jury. See State v. Hunt, 115 N.J. 330 , 380, 558 A.2d 1259 (1989) (citing Lowenfield v. Phelps, 484 U.S. 231, 240, 108 S.Ct. 546, 552, 98 L.Ed.2d 568, 579 (1988) (approving of an inquiry about whether "further deliberations might assist them" after jury indicated an inability to reach a unanimous verdict)). In a capital case, this inquiry is crucial because a "stalemate" is a legitimate and acceptable conclusion of jury deliberations. In this case, the jury did not initially come to the court with a suggestion of impasse, but with a complaint about Juror Number Seven's introduction of extraneous information into deliberations. However, on learning the extent that the jury was divided at the very late stage of the deliberative process, the court had an obligation to inquire further. The court failed to ascertain whether the jury had reached a non-unanimous verdict prior to the misconduct or whether the misconduct was capable of tainting further jury deliberations.

In analyzing whether the initial jury had reached a non-unanimous verdict, the majority's opinion poses a foundational question drawn out of Ramseur and subsequent cases: Would further deliberations help the jury reach an uncoerced unanimous decision. Ante at 258-60, 680 A.2d at 658-59. In this case, the trial court had the unusual advantage of knowing who the holdout was. As a result, its inquiry into whether the power of persuasive argument had run its course need not have been overly abstract. Indeed, several comments made by Juror Number Seven and other members of the jury quite directly addressed the question of whether Juror Seven, the holdout, had reached an "honest conviction as to the weight or effect of the evidence" that further deliberations were unlikely to alter. State v. Czachor, 82 N.J. 392 , 405 n. 4, 413 A.2d 593 (1980) (recommending use of ABA Model Charge in criminal trials).

It is not necessary to establish, as the majority suggests, that Juror Number Seven would never have changed his mind. Such a burden could not be satisfied. More to the point, that burden need not be satisfied. Rather, we recognized the appropriate standard in Ramseur and subsequent cases: whether to justify the discontinuance of deliberations there was "'a substantial likelihood,' that a non-unanimous verdict would have been returned. . . ." Hunt, supra, 115 N.J. at 381, 558 A.2d 1259 (quoting Ramseur, supra, 106 N.J. at 314, 524 A.2d 188). If there was, "we cannot require defendant to run the risk that a different jury might decide he should die." Ibid.

The majority's attempts to reconcile and distinguish Ramseur are not convincing. First the majority argues that the trial court did not abuse its discretion in determining that the original jury did not deliberate for a reasonable amount of time prior to reaching a deadlock. Ante at 258-59, 680 A.2d at 658-59. The majority points to the complexity of a trial involving the c(4)(f) (avoid apprehension) factor and the c(4)(g) (murder during the course of a robbery) factor. The majority also notes that the jury had to make a determination on defendant's mental state. However, the proofs in this trial were not as complex as suggested by the majority. Both the c(4)(f) and the c(4)(g) factors involved evaluation of the same evidence. See discussion, infra at 285-89, 680 A.2d at 672-74. Also, it was necessary to determine defendant's mental state in deciding the c(4)(f) factor. Finally, when Juror Number Seven interjected the extra-record information, the jury had deliberated six and one-half hours, inferentially an adequate time in view of the fact that the reconstituted jury, presumably starting its deliberations anew, reached a final verdict in four and one-half hours.

The majority also relies on the fact that, unlike in Ramseur, there was no note or direct admission of deadlock. Rather, the majority argues that the evidence was suggestive of continuing, intense debate. Ante at 258-60, 680 A.2d at 658-59. The distinction is unavailing. Although the issue of deadlock did not arise from an independent communication from the jury, there was actually more compelling evidence of deadlock here than in Ramseur. Aside from Juror Number Seven's plain statement in favor of imposing a life sentence, the strongest evidence that Juror Seven was emerging as the lone voice in support of a life sentence appears in the foreman's comment that Juror Number Seven "has a point of view that deserves to be on the jury, and . . . I'm afraid . . . that he might be removed . . . [and] that his point of view gets lost." The foreman's comment strongly suggests that the jury was prepared and ready to agree to disagree to a non-unanimous verdict authorized under the capital-punishment statute.

The record is replete with additional evidence that deliberations had proceeded very far. The fact that jurors were still debating the issue is not necessarily indicative of the fact that minds were not made up or a decision was not reached. If Juror Seven was the only vote for life, and he was continuing to attempt to persuade the others to change their minds, serious debate would consequently follow.

Comments made by other jurors suggest that Juror Seven had, like Juror Number Two, made his decision, and would not be persuaded by any further argument to change his mind. Juror Number Ten, for example, stated that he had the impression that Juror Seven "pretty much had his mind made up" at an early stage of the discussions. The record demonstrates that jurors were "taking positions," that at least one juror other than Juror Seven had made up his mind, and that there was frustration at the position Juror Seven was taking. Therefore, there is a substantial likelihood that further deliberations would not have helped this jury reach uncoerced unanimity.

The jury had demonstrated a substantial likelihood of a non-unanimous verdict. The court was obligated to further investigate this potential. The absence of investigation in combination with the strong evidence that Juror Number Seven could have successfully remained on the jury, dictate that this Court must respect the potential for a life sentence as it did in Ramseur. Defendant may not, on remand, be exposed to the death penalty.

II

The Court in this case endorses an expansive interpretation and application of the c(4)(f) aggravating factor. That factor enables a jury to impose a death sentence if the defendant committed the murder to avoid apprehension. The Court's understanding of what may be encompassed by this aggravating factor has the potential to convert any felony murder into capital murder.

This is not the first time that I have written about the dangers inherent in a broad reading of c(4)(f). See, e.g., Hightower I, supra, 120 N.J. at 434-38, 577 A.2d 99 (Handler, J., dissenting). This case demonstrates even more clearly the dangers previously noted. I reassert the need to clarify and limit the standard governing the admissibility and use of evidence offered to prove c(4)(f), including restriction of its application to prior offenses unrelated to the murder itself. Id. at 437, 577 A.2d 99 (citing State v. Monturi, 195 N.J. Super. 317, 327, 478 A.2d 1266 (Law Div. 1984) ("[T]he motive behind the murder must be the concealment of another prior offense, and the proofs must be limited to that issue.")). Most troubling, the presently vague and malleable standard for raising the c(4)(f) factor always permits the evidence supporting the felony murder factor, c(4)(g), to support the c(4)(f) factor. In other words, the c(4)(f) factor automatically duplicates the c(4)(g) factor. Id. at 437-38, 577 A.2d 99. At a minimum, in order to demonstrate sufficient evidence, the State must proffer factual evidence that is independently significant to each factor. Without any further statutory or judicial guidance to channel prosecutorial and jury discretion, however, the c(4)(f) factor clearly fails to satisfy constitutional mandates under the Eighth and Fourteenth Amendments.

The majority's finding of sufficient evidence to support the c(4)(f) aggravating factor substantially reduces the State's evidentiary burden, rendering the reasonable doubt standard worthless in its singular function to assure the principled imposition of the death sentence. The general and attenuated evidence, relied on by the State in arguing the factor, can be found in any felony murder. Yet, clearly not every felony murder can be treated as a capital case. More troubling, the indirect and equivocal circumstantial evidence can be shaped and directed, with equal force and conviction, to support different motives for murder. In finding the reasonable doubt standard satisfied by such amorphous evidence, the majority dilutes the reasonable doubt standard to equate with no more than the preponderance of the evidence standard.

A.

We have been over the facts many times: State witness Christopher Forston testified that soon after the murder, defendant described the crime to Forston. Defendant went to the Cumberland Farms store in Willingboro, New Jersey, just before noon on July 7, 1985, intending to commit a robbery. Defendant lingered in the store for a brief period until all of the customers left. Defendant then approached the sales counter, placed a box of Pampers on the counter, and asked for a carton of cigarettes. When the sales clerk, Cynthia Barlieb, turned away to get the cigarettes, defendant turned the store's open sign to closed and pulled out his gun. He demanded that Ms. Barlieb open the register. When she did not, he shot her in the chest. Ms. Barlieb fell, but then got back up. Defendant again told her to open the cash register. Ms. Barlieb again refused and "hollered or made a movement." Defendant shot her again, this time in the neck. Defendant then jumped over the counter and attempted to open the register himself. Unable to do so, he ordered Ms. Barlieb to show him how the machine opened. Forston testified that "she wouldn't show him so he kept messing with it and he . . . started getting mad, and he jammed the cash register and he shot her again." The third shot hit Ms. Barlieb in the head. Forston testified further, that in response to Forston's question as to why defendant had killed Ms. Barlieb, defendant answered that he had done so because "the old bitch wouldn't cooperate."

The record reveals little credible evidence supporting the argument that in killing Barlieb defendant was motivated by a fear of future identification. First, defendant did not know Cynthia Barlieb. Moreover, while defendant's parents resided several blocks from the convenience store, defendant spent most of his time in Texas and Philadelphia. Nor did defendant attempt to conceal his identity from Ms. Barlieb or other customers upon entering the store. For example, he did not wear a mask and he entered the store at a time when other customers were present. Additionally, there was no extended period of time prior to the shooting during which defendant and the victim were in contact.

The majority points to six aspects of the evidence in finding sufficient proof of the c(4)(f) factor. Two of the factors detail where defendant shot the victim — i.e., the heart and the head. These facts imply nothing but an intent to kill. The majority also points out that the second shot was fired after the victim hollered. The first and third shots were not, however, fired in response to the victim's cries. Moreover, that the victim yelled after being shot is not proof of defendant's motive in shooting her. The majority also argues that the time of day supports a finding of the factor. The majority could just as easily point to a nighttime robbery as evidence of a murder with the intent to avoid apprehension. The fact that the robbery occurred during the day or the night, however, does not even arguably reveal defendant's intent. Finally, the majority asserts that defendant waited for the customers in the store to leave and turned the sign in the store window to the closed position. These actions are more readily explained as a desire to avoid a disturbance during the course of the robbery, and thus an intent to rob.

Significantly, there is compelling evidence that defendant killed Ms. Barlieb for another reason: she was not cooperating. Each shot occurred after a repeated request to gain the assistance of the clerk. Only when she refused to cooperate did defendant shoot her. On the other hand, the State's basis for the c(4)(f) factor, that defendant shot Ms. Barlieb to quiet her or to ensure that she would not later identify him, is not supported by anything more than speculation. Especially where a strong and clear alternative motive exists on the record, this Court must demand a showing of evidence supporting the c(4)(f) factor, whether circumstantial or direct, that rises above the level of speculation.

B.

Although the United States Supreme Court has found that death is a constitutional punishment under the Eighth Amendment, the Court emphasized that "[death] is an extreme sanction, suitable to the most extreme of crimes." Gregg v. Georgia, 428 U.S. 153, 187, 96 S.Ct. 2909, 2932, 49 L.Ed.2d 859, 882 (1976). "Because of the uniqueness of the death penalty, . . . it [may] not be imposed under sentencing procedures that create a substantial risk that it [will] be inflicted in an arbitrary and capricious manner." Id. at 188, 96 S.Ct. at 2932, 49 L.Ed.2d at 883 (citing Furman v. Georgia, 408 U.S. 238, 313, 92 S.Ct. 2726, 2764, 33 L.Ed.2d 346, 392 (1972) (White, J., concurring)). Recognizing the inevitable inexperience of the jury in performing sentencing determinations as well as the complexity of the task, id. at 192, 96 S.Ct. at 2934, 49 L.Ed.2d at 885, the Supreme Court emphasized the need for a structured and "meaningful basis for distinguishing the few cases in which [death] is imposed from the many cases in which it is not." Id. at 188, 96 S.Ct. at 2932, 49 L.Ed.2d at 883 (citing Furman, supra, 408 U.S. at 313, 92 S.Ct. at 2764, 33 L.Ed.2d at 392 (White, J., concurring)).

Thus, "[t]o pass constitutional muster, a capital sentencing scheme must genuinely narrow the class of persons eligible for the death penalty and must reasonably justify the imposition of a more severe sentence on the defendant compared to others found guilty of murder." Lowenfield v. Phelps, supra, 484 U.S. at 244, 108 S.Ct. at 554, 98 L.Ed.2d at 581 (quotation and citation omitted). One method of guiding the jury's decision making process, approved by the Supreme Court in Gregg, and adopted by the New Jersey Legislature and this Court in Ramseur, is through statutory aggravating and mitigating circumstances focusing the jury on the character of the defendant and the nature of the crime. Gregg, supra, 428 U.S. at 192-95, 96 S.Ct. at 2934-35, 49 L.Ed.2d at 885 ; Ramseur, supra, 106 N.J. at 197, 524 A.2d 188. This sentencing structure "pass[es] constitutional muster, [because it]: limit[s] imposition of the penalty to what is assumed to be the small group for which is it appropriate . . . and ensure[s] that the limited class selected for the penalty is chosen with rationality and consistency." Ramseur, supra, 106 N.J. at 183, 524 A.2d 188 (citations omitted).

Under our capital sentencing scheme, then, aggravating factors provide constitutionally necessary guidance for jury and prosecutorial discretion in determining for whom the penalty of death is the appropriate sanction. Because of the vital limiting role of the aggravating factors, the factors themselves must be sufficiently clear and precise. A vague or overly broad factor simply becomes a mask for unfettered sentencing discretion; it cannot effectively control, direct, or limit that discretion. The United States Supreme Court instructs that the test for the breadth of an aggravating factor is not whether the factor, on its face, can be interpreted too broadly, but whether the factor, as interpreted by the supreme court of the state, is sufficiently narrow. Gregg, supra, 428 U.S. at 201, 96 S.Ct. at 2938, 49 L.Ed.2d at 890 (noting that the "outrageously or wantonly vile" aggravating factor can be construed to incorporate any murder, but finding no evidence that the Georgia Supreme Court would adopt such a broad interpretation).

The c(4)(f) factor, by its terms and as interpreted by this Court, focuses on the defendant's purpose in killing. In State v. Martini, 131 N.J. 176 , 619 A.2d 1208 (1993) ( Martini I), the Court explained: "[t]he key to finding factor c(4)(f) is that the defendant intended to eliminate a potential witness to his crimes." Id. at 281, 619 A.2d 1208; see Hightower I, supra, 120 N.J. at 420-21, 577 A.2d 99 (finding that "'[t]he gravamen of [the (4)(f)] factor is the silencing of potential witnesses, which could be for the underlying crime being committed or for a crime committed at another time'") (quoting State v. James Moore, 207 N.J. Super. 561 , 569, 504 A.2d 804 (Law Div. 1985)).

As noted by the majority, the Court has previously embraced expansive interpretations of the factor. First, the Court has held that the factor may include cases in which the crime that the defendant seeks to conceal is contemporaneous with the murder itself. Hightower I, supra, 120 N.J. at 420, 577 A.2d 99. Second, the Court allowed application of the factor where the purpose of eliminating a witness was but one of several motives inducing the defendant to kill. Martini I, supra, 131 N.J. at 282, 619 A.2d 1208 (reasoning that "[n]othing in the language of the factor indicates that it was meant to apply only to those murders undertaken solely for the purpose of eliminating a witness"). Finally, although the Court has stated the trial court must "warn the jury not to double-count the evidence," it acknowledged that "the same evidence can support more than one aggravating circumstance." Hightower I, supra, 120 N.J. at 422, 577 A.2d 99.

The concern over double-counting of evidence is highlighted where, as here, the c(4)(g) (murder during a felony) and c(4)(f) aggravating factors are both charged. Given that criminals generally want to escape detection for their crimes, Martini I, supra, 131 N.J. at 283, 619 A.2d 1208, every defendant who commits murder during a rape, robbery, or other felony listed in c(4)(g), will automatically also be liable under an overly broad interpretation of c(4)(f). Id. at 281, 284, 619 A.2d 1208; Hightower I, supra, 120 N.J. at 422, 426, 577 A.2d 99. This Court has, however, determined that automatic duplication is not acceptable. See Martini I, supra, 131 N.J. at 281, 619 A.2d 1208 ("our refusal to construe c(4)(f) narrowly as requested by defendant does not mean that in all felony murders that circumstance will duplicate c(4)(g)"). "[T]he mere fact of a death is not enough to invoke this factor." Hightower I, supra, 120 N.J. at 422, 577 A.2d 99 (quotation omitted). Today's holding turns the rule against double-counting into an empty admonition.

To ensure that the c(4)(f) factor does not in fact duplicate c(4)(g), each factor must potentially cover some situation not covered by the other. Otherwise, the c(4)(g) factor would constitute a completely contained subset of the c(4)(f) factor, such that every c(4)(g) case is also a c(4)(f) case ( i.e., every murder during a felony could also be considered a murder to eliminate a witness to the felony). If duplication of charges or aggravating factors is avoided only because each factor can cover some situations not covered by the other, that distinction between the factors must manifest itself in a distinction relating to the evidence relevant to the factors. If automatic duplication were accepted, then the c(4)(f) factor would have no potential to further narrow the death-eligible class of offenders. Rather, it would be present in every c(4)(g) case, and serve only to emphasize the blameworthiness of a felony murder offender rather than to distinguish between felony murder offenders.

The Court can preserve the c(4)(f) factor as distinct and not duplicative of the c(4)(g) factor. As noted above, the c(4)(f) factor exists when a defendant's decision to kill the victim arises, at least in part, out of a desire to escape prosecution. In other words, the c(4)(f) factor should not necessarily exist whenever a defendant both desires to escape prosecution (as all do) and murders during the course of another felony (the c(4)(g) factor). Rather, it must further be shown that the defendant killed the victim because the defendant desired to escape prosecution.

The majority's unsound and unexplained departure from our commitment to fairness and consistency in death-penalty convictions is highlighted by comparing the present adoption of a broad reading of the c(4)(f) factor to the extreme measures this Court found necessary in Ramseur to limit and confine the c(4)(c) factor, ensuring that it provided sufficient guidance in determining death eligibility. In Ramseur, the Court emphasized its commitment to "the goals of consistency and reliability in the administration of the capital punishment" despite the United States Supreme Court's apparent retrenchment in those areas. 106 N.J. at 190, 524 A.2d 188. The Court emphasized that "[i]n the context of the death penalty, where the demands for fairness and accuracy are heightened, the principles of consistency and reliability rise to constitutional dimension." Ibid.

The Ramseur Court faced a challenge to the c(4)(c) factor ("[t]he murder was outrageously or wantonly vile, horrible or inhuman in that it involved torture, depravity of mind, or an aggravated battery to the victim") similar to the present challenge to the c(4)(f) factor. There, the defendant alleged that the c(4)(c) factor was facially unconstitutional because it was so broad and vague it could not be construed in a manner that sufficiently guided the jury's discretion. The Court explained that the vagueness resulting from the construction of the aggravating factor "allows arbitrary and discriminatory enforcement of the laws by police, judges and juries." Id. at 201 n. 27, 524 A.2d 188. The result was to render "the jury's functioning unpredictable. In this context, a vague sentencing guideline, by rendering the determination more inscrutable to judicial review, increases the chance for concealed prejudice in sentencing." Ibid. Recognizing both its "power and obligation to narrow imprecise statutory language in order to render it constitutional," id. at 200, 524 A.2d 188 (citations omitted), the Court found that "the narrowing [of the c(4)(c) factor] is essential to satisfy the requirement of Gregg that the discretion of the jury be adequately controlled and the requirement of the fourteenth amendment that criminal laws not be vague." Id. at 200-01, 524 A.2d 188.

Therefore, the Court painstakingly delineated each term of the factor, including "torture" and "depravity of mind." The Court also defined and emphasized the need for both purposeful and knowing commission of the factor, providing clear examples of boundaries that trial courts were to respect. Wary of the fact that intent often must be demonstrated with circumstantial evidence, and that mere speculation could not provide a finding of a factor of such import beyond a reasonable doubt, the Court held that "pain in fact" is required to establish the existence of the factor. Therefore, the Court held that the death penalty could not be imposed "where no pain was suffered despite the murderer's intent to inflict it . . . [because the State could not tolerate] discretionary findings and death sentences based on the slimmest of evidence." Id. at 208, 524 A.2d 188. "[T]he extreme physical or mental suffering must be precisely what defendant wanted to occur in addition to death." Id. at 208-09, 524 A.2d 188. The Court further limited the factor to require both a knowing and purposeful state of mind, holding that the factor could not be found even if a defendant, although "'practically certain' that extreme physical or mental suffering would occur, did not have that result as his 'conscious object.'" Id. at 209 n. 35, 524 A.2d 188. Thus, the Court "provide[d] further assurance that this aggravating factor will apply only to the most culpable murder." Ibid.

Yet, in contrast, the Court today indifferently accepts a standard for the avoid apprehension factor that allows that factor to be raised in all felony murder cases. Some restrictive applications may be imposed by individual courts, prosecutors, or juries in given felony murder cases. But that type of unprincipled application and unguided discretion "allows arbitrary and discriminating enforcement of the laws by police, judges and juries." Id. at 201 n. 27, 524 A.2d 188. A limitation that is random, unstructured, and unprincipled is not sufficient, in a system demanding guided discretion, to ensure constitutional administration of the gravest sanction for the most serious of crimes.

C.

Study of previous rulings by this Court regarding the sufficiency of the evidence supporting the c(4)(f) factor suggests that the Court was aware of the need to establish a causal connection between the murder and the desire to escape prosecution. Thus, the Court focused on evidence tending specifically to demonstrate the formation of a pre-murder intent to eliminate a witness to some underlying felony. Credible and persuasive evidence establishing the c(4)(f) factor has included admissions of motive as well as proof of a prior relationship between the defendant and victim or an extended period of exposure, prior to the murder, during which the victim could study the defendant's appearance for subsequent identification.

For example, in past cases affirming the c(4)(f) factor, the Court has relied on the confession of the defendant that the purpose for the murder was to escape prosecution for that or some other crime. See, e.g., State v. DiFrisco, 137 N.J. 434 , 500-02, 645 A.2d 734 (1994) ( DiFrisco II) (noting defendant's statement that he killed victim to eliminate witness to procurer's criminal activity); Martini I, supra, 131 N.J. at 280, 619 A.2d 1208 (noting that defendant informed victim's wife that defendant was wanted by the authorities for a prior murder and later confessed that he murdered victim because "I thought he was going to run away"); State v. Rose, 112 N.J. 454, 531, 548 A.2d 1058 (1988) (affirming factor where only evidence of defendant's motive was his own statement that he panicked and did not want to be caught with shotgun). Additionally, the Court in Martini I affirmed a finding of sufficient evidence to support the c(4)(f) factor, finding "most persuasive the fact that the victim had been acquainted with defendant and was in his presence for an extended period of time during which defendant's identity was not disguised." 131 N.J. at 283, 619 A.2d 1208. The Court further noted that "no apparent purpose was served by [the] murder except for defendant's escape from detection." Id. at 284, 619 A.2d 1208. In respect of evidence showing that defendant attempted to elude police, the Court stated:

The steps that defendant took to avoid apprehension, including his eluding police after picking up the ransom money, also shed light on his motive for the murder. True, as defendant argues, everyone who commits a felony seeks to avoid detection. Standing alone, those circumstances may not support application of the factor. However, when a jury views those circumstances along with the victim's familiarity with defendant, it could reasonably infer that the defendant firmly intended to escape and that the elimination of Flax was essential to the fulfillment of that objective.

[ Ibid.]

This Court has identified the standard governing submission of aggravating factors to the jury. See Hightower I, supra, 120 N.J. at 420, 577 A.2d 99 (holding that the "State must provide evidence from which jury could rationally conclude beyond a reasonable doubt that" the aggravating factor is present); see also Martini I, supra, 131 N.J. at 282-83, 619 A.2d 1208 (holding that "aggravating factor c(4)(f) may be submitted to the jury when the State has produced sufficient evidence on which a jury can reasonably conclude that at least one of the motives of the defendant in killing his or her victim was to eliminate a witness or avoid subsequent apprehension and prosecution for criminal acts"). Yet, in this case, the majority neither acknowledges the absence of factual support similar to that found in prior cases affirming the c(4)(f) factor, nor attempts to account for the Court's dramatic expansion of the factor in the present case.

The need for further narrowing is additionally supported by decisions of other jurisdictions. For example, many courts refuse to allow evidence that the victim was able to observe the defendant, in and of itself, to support the avoid apprehension factor. See, e.g., People v. Bigelow, 37 Cal.3d 731, 209 Cal.Rptr. 328, 331-32, 691 P.2d 994, 997-98 (1984) (finding argument that murder was committed to avoid apprehension was "totally speculative" where defendant hitchhiked a ride with victim, took victim to a remote location, shot victim, and drove away with his car); Hansbrough v. State, 509 So.2d 1081, 1086 (Fla. 1987) (holding that "the mere fact that the victim might have been able to identify her assailant is not sufficient to support finding this factor"); State v. Williams, 304 N.C. 394, 284 S.E.2d 437, 455-56 (1981) (requiring direct evidence of motive to avoid apprehension and finding insufficient evidence of motive where defendant killed convenience store clerk and attempted to escape in an inconspicuous fashion), cert. denied, 456 U.S. 932, 102 S.Ct. 1985, 72 L.Ed.2d 450 (1982); Commonwealth v. Daniels, 537 Pa. 464, 644 A.2d 1175, 1179 (1994) (requiring direct evidence to support the avoid apprehension factor and holding that the State's burden "will not be met by simply showing that an individual who witnessed a murder or other felony committed by a defendant was also killed by the defendant"); State v. Branam, 855 S.W.2d 563, 570 (Tenn. 1993) (holding that even though victim knew defendant and victim sounded her car horn after the first shot, there was no evidence beyond speculation to support the contention that the murder was committed to avoid apprehension for the robbery).

Courts in other states frequently require the production of very focused evidence of intent to eliminate a witness in order to justify submission of such a factor to the jury. See, e.g., Hannon v. State, 638 So.2d 39, 43-44 (Fla. 1994) (holding that "when a murder victim is not a police officer, '[p]roof of the requisite intent to avoid arrest and detection must be very strong.'") (quotation omitted), cert. denied, ___ U.S. ___, 115 S.Ct. 1118, 130 L.Ed.2d 1081 (1995). Exemplary of the specific kind of evidence offered in support of the analogous aggravating factor in other states is evidence that a defendant removed the victim to a remote location before the killing, see, e.g., Preston v. State, 607 So.2d 404, 409 (Fla. 1992), cert. denied, 507 U.S. 999, 113 S.Ct. 1619, 123 L.Ed.2d 178 (1993); Munson v. State, 758 P.2d 324, 335 (Okla.Cr. 1988), cert. denied, 488 U.S. 1019, 109 S.Ct. 820, 102 L.Ed.2d 809 (1989); that a defendant killed or attempted to kill more than one person who witnessed an underlying offense, see, e.g., Pickens v. State, 292 Ark. 362 , 730 S.W.2d 230, 234, cert. denied, 484 U.S. 917, 108 S.Ct. 269, 98 L.Ed.2d 226 (1987); Hannon, supra, 638 So.2d at 43-44; State v. Simants, 197 Neb. 549, 250 N.W.2d 881, 891, cert. denied, 434 U.S. 878, 98 S.Ct. 231, 54 L.Ed.2d 158 (1977); State v. Green, 321 N.C. 594, 365 S.E.2d 587, 595-96, cert. denied, 488 U.S. 900, 109 S.Ct. 247, 102 L.Ed.2d 235 (1988); that a defendant was acquainted with the victim, see, e.g., Wainwright v. State, 302 Ark. 371 , 790 S.W.2d 420, 427 (1990), cert. denied, 499 U.S. 913, 111 S.Ct. 1123, 113 L.Ed.2d 231 (1991); Riley v. State, supra, 366 So.2d at 22; State v. Palmer, 224 Neb. 282, 399 N.W.2d 706, 726 (1986), cert. denied, 484 U.S. 872, 108 S.Ct. 206, 98 L.Ed.2d 157 (1987); Roberts v. State, 868 P.2d 712, 722 (Okla.Cr.), cert. denied, ___ U.S. ___, 115 S.Ct. 158, 130 L.Ed.2d 96 (1994); or that the victim attempted to escape during the crime and was killed at the time of the attempted escape. See, e.g., Young v. State, 579 So.2d 721, 724 (Fla. 1991), cert. denied, 502 U.S. 1105, 112 S.Ct. 1198, 117 L.Ed.2d 438 (1992); People v. Williams, 97 Ill.2d 252, 73 Ill.Dec. 360, 369, 454 N.E.2d 220, 229 (1983), cert. denied, 466 U.S. 981, 104 S.Ct. 2364, 80 L.Ed.2d 836 (1984); State v. Smith, 868 S.W.2d 561, 580 (Tenn. 1993), cert. denied, ___ U.S. ___, 115 S.Ct. 417, 130 L.Ed.2d 333 (1994).

D.

The majority's error stems from its deficient interpretational definition of the c(4)(f) factor. After reciting the statutory definition of the factor, the majority notes that we have accepted a broad interpretation of c(4)(f), requiring that the jury need find only that "one of the motives of the defendant in killing his or her victim was to eliminate a witness or avoid subsequent apprehension and prosecution for criminal acts." Ante 267-68, 680 A.2d at 663.

The majority's definition, however, is woefully incomplete. The majority's interpretation of the avoid apprehension factor violates the Eighth Amendment prohibition against cruel and unusual punishment because, when accompanied by the felony murder factor, the avoid-apprehension factor fails to provide any further basis for distinguishing deathworthy offenders as instructed under the Furman doctrine. The "narrowing" construction adopted by the majority, however, is blatantly ineffective and does not even provide a facially objective means of isolating the c(4)(f) factor. The majority's interpretation of the factor thereby tolerates the automatic duplication that it rejected in Martini I. Consider, the Court recognized that it is almost always possible to argue that a defendant who commits a felony also desires to avoid apprehension. Moreover, the Court will allow a finding of the c(4)(f) factor even where the evidence demonstrates an alternative motive and no specific evidence demonstrates that the murder was committed for the purpose of avoiding apprehension. Thus, the factor is so malleable that it can be found in any case. Without any guidelines, there is no basis for the State to make rational and consistent determinations in charging the factor, and no basis for a deliberating jury to accept or reject the State's argument. If the c(4)(f) factor can be raised in all c(4)(g) cases, but the State does not raise it in every case, the death selection process is under the arbitrary control of the State.

This unintentional delegation of discretion has serious consequences. First, a prosecutor charging the c(4)(f) factor can introduce prejudicial evidence that would not otherwise be admissible, except to establish, by circumstantial evidence, that the motive was to avoid apprehension. Further, merely charging a second aggravating factor undoubtably increases the likelihood of death. State v. Harris, A-108-95, Proportionality Review Study, Tables 8 9. A jury is more likely to punish a defendant charged with numerous aggravating factors. Additionally, because there are no clear guidelines or established rules about when c(4)(f) applies, the prosecutorial discretion also exposes the defendant to increased jury discretion. The bias of the additional factor is clearly weighted towards death. Without further limitation, the factor "allow[s] for standardless and arbitrary application of the law by prosecutors and juries." Ramseur, supra, 106 N.J. at 198, 524 A.2d 188.

III

I join in the Court's decision to reverse Hightower's death sentence because the trial court abused its discretion in removing Juror Number Seven from the jury and because deliberations had gone too far for substitution and reconstitution of the jury. However, this Court's determination that the trial court should have granted a mistrial is wrong. Under the circumstances, defendant was denied the realistic opportunity for a non-unanimous verdict. Defendant should not again be exposed to a death sentence and subjected to another penalty trial. Additionally, the Court's expansive interpretation of the c(4)(f) aggravating factor and its tolerance of duplicative, overlapping evidence sanctions discretion that is insufficiently guided, and will inevitably result in the arbitrary, discriminatory, and unconstitutional application of the death penalty.

Therefore, I respectfully dissent.

For reversal and remandment — Justices POLLOCK, GARIBALDI, STEIN and COLEMAN — 4.

For concurrence in part, dissent in part — Justices HANDLER and O'HERN — 2.