I respectfully dissent.
I cannot agree with the reasoning of part II of the majority opinion, which dispenses with the necessity for consistency in the votes of jurors whose special verdicts determine the comparative negligence of two or more parties and apportion liability among them. As the majority acknowledges, the legal requirement that at least three-fourths of a jury must agree to a verdict in a civil case (see Cal. Const., art. I, § 16; Code Civ. Proc., § 618) has long, and uniformly, been interpreted to require that at least nine of the same jurors must agree upon all elements of the ultimate general verdict. (Ante, p. 766; and cases cited.) Without offering any rationale for their distinction, however, my colleagues depart from that requirement when those underlying elements take the form of special verdicts, holding that so long as nine jurors find each party liable, “any nine jurors” may apportion damages among them. (Ante, p. 768.) In my view, the well established rule which requires consistency in the votes of jurors to support a general verdict is equally applicable where the ultimate verdict rests upon specific questions denominated as “special verdicts.”
In the present case, when the jury was polled on November 6, 1980, it became apparent that although each special verdict was supported by nine votes, two of the nine jurors who apportioned liability for petitioner’s damages among petitioner, petitioner’s employer and real party in interest (real party), had found either that real party had not been negligent at all or that petitioner had not been negligent at all. Thus, juror number 1, who found that plaintiff (petitioner) was not negligent, nonetheless attributed to plaintiff 25 percent of the liability for his damages. Similarly, juror number 7 attributed 45 percent of the liability for the damages to defendant (real party), although that juror specifically found that defendant was not negligent at all. In fact, only seven of the jurors ultimately assessing and apportioning damages among all three *770parties found all three parties to be negligent and the negligence of all three to be the proximate cause of petitioner’s damages. (While actionable negligence traditionally involves demonstration of breach of a duty of care proximately causing actual damages [see Prosser, Torts (4th ed. 1971) § 30, p. 143], I here adopt the terminology of the special interrogatories used in this case,! which obviously used the term “negligence” to refer only to the breach of the duty of care, and inquired separately as to whether such “negligence” proximately caused plaintiff’s damages.)
The majority urges that ithe blatant inconsistency between a juror’s factual finding that a party is not negligent, and his determination that such nonnegligent party is ■ nonetheless liable to pay damages for the negligent infliction of injuries, need not be considered illogical if it can be assumed that a juror dissenting from a finding of liability will accept the majority’s finding on that issue before proceeding to consider the apportionment of that liability among the parties. (Ante, p. 768.) What the majority does not suggest, however, is why such reasoning should be adopted in the context of a case decided upon special verdicts when it has never been applied to a case decided by general verdict in this state.
Nor does United Farm Workers of America v. Superior Court (1980) 111 Cal.App.3d 1009 [169 Cal.Rptr. 94], provide persuasive support for the majority’s holding. In United Farm Workers the Court of Appeal found error in the trial court’s declaration of a mistrial for a defective verdict on alternative grounds. Procedurally, it relied upon Henrioulle v. Marin Ventures, Inc. (1978) 20 Cal.3d 512, 521 [143 Cal.Rptr. 247, 573 P.2d 465], to conclude that defendant had waived any defect in the jury verdict by failing to object thereto and to request further deliberation thereon before the jury was discharged. (111 Cal. App.3d, at pp. 1021-1022.) However, it also addressed the merits of defendant’s claim that the júry verdict was defective, declaring: “We decide that under the circumstances in this case it is not fatal to the verdict that the same nine jurors did not find both defendant and plaintiff negligent as long as nine jurors did so find, and that the award of damages was properly determined.” (Id., at p. 1021.)
The United Farm Workers' court observed that the facts supporting a finding that one party is negligent may have no necessary bearing on a finding that another party is negligent. While that observation is undoubtedly correct, it has little bearing on the crucial issue before us. Although applying the generally acceptable principle that special ver*771diets which are not logically interdependent need not be approved by the same nine jurors, it stops short of the question fairly raised here: must there be consistency between a juror’s apportionment of liability among two or more parties and his finding that each of those parties was culpable? Neither by the language of its holding which we have quoted (see also 111 Cal.App.3d, at p. 1019), nor by the rationale therefor which we have discussed, did the United Farm Workers court purport to dispense with the requirement of consistency in the votes of jurors on special verdicts whose reasonableness depends upon consistency.
An examination of the jury poll in United Farm Workers, however, does reveal a logical inconsistency which the court did not analyze, and which would appear to violate the rationale of its decision. In that case, of the 11 jurors who attributed to plaintiff 90 percent of the responsibility for his $70,000 in damages, only 8 found that plaintiff’s negligence was the proximate cause of any damages at all. (111 Cal.App.3d, at p. 1011.) It would seem contrary to the fundamental purpose of our comparative negligence system — a purpose which we have identified as being “to assign responsibility and liability for damage in direct proportion to the amount of negligence of each of the parties” (Li v. Yellow Cab Co. (1975) 13 Cal.3d 804, 829 [119 Cal.Rptr. 858, 532 P.2d 1226, 78 A.L.R.3d 393]) — to sustain a verdict imposing 90 percent of liability on a party by counting the vote of a juror who has concluded either that such party was not negligent at all or that his negligence was not the proximate cause of any damages. In addition, of course, such inconsistent conclusions simply are not reasonable, and reason is the foundation upon which jury deliberation rests. (See Naumburg v. Wagner (1970) 81 N.M. 242, 243 [465 P.2d 521, 522] [disqualifying any juror who is “guilty of irreconcilable inconsistencies or material contradictions when his votes on all issues are considered”].)
Nor does the requirement of consistency between a juror’s findings of negligence, proximate cause, and damages, on the one hand, and liability for those damages, on the other, necessarily require, as the majority suggests (ante, p. 768), that only those jurors initially finding each of the parties liable may deliberate upon the apportionment of that liability among them. While the logical progression would normally appear to be from findings of actionable negligence to apportionment of liability therefor, it is apparent that a juror may change her or his mind on one or more special verdicts at any time before the jury verdict is rendered. That factor alone would appear to refute the wisdom of imposing *772upon the deliberative process an artificial restriction barring each juror from fully considering and reconsidering each issue. As the majority implicitly acknowledges by its holding, a tentative vote is not cast in bronze; and so long as the ultimate product of a juror’s deliberation is consistent, there is no reason to interfere with the process of “decisions and revisions which a minute [may] reverse.” (T. S. Eliot, “The Love Song of J. Alfred Prufrock.”)
In my view, the jury instructions given here adequately guided the jurors in the appropriate procedure. Apparently without objection, the jury was instructed generally in accordance with BAJI No. 15.32 (1977 ed.) as follows: “Each of you should deliberate and vote on each issue to be decided. However, before you may return a verdict to the court, at least nine or more identical jurors must agree on the verdict in its final and complete form, so that each of those nine or more may be able to state truthfully that the verdict is his or hers.” It was also instructed more specifically in accordance with BAJI No. 15.50 (1977 ed.) that “nine or more identical jutors . . . [must agree] upon each answer required by . . . [the] directions on the special verdict form, so that each of those nine or more may , be able to state truthfully that every answer is his or hers .. . . ” Nothing in the instructions purported to bar jurors from full participation in the deliberative process at any point before a final verdict was rendered.
Further, while the adoption of a rule abandoning the necessity of consistency in a juror’s special! verdicts in this context may expedite such trials, it is not likely to be less confusing to the jury, as the majority contends, or more productive of justice. Rather, informing jurors, as the present instructions do, thát they must be consistent in attributing responsibility for the payment of damages to parties whom they find to be culpable would appear to (lave the dual virtues of simplicity and common sense, while effectuating the underlying principle of responsibility in proportion to fault. Where the rule is both logical and clear, the likelihood of its implemention ¡ would seem greater.
Finally, as a practical matter, it does not seem to me realistic to assume that a juror who concludes that a party is not culpable would be able conscientiously to apportion financial responsibility to that party. His perception of a legal compulsion upon him to affix some responsibility upon a party whom he concludes is not responsible at all is more likely to cause that juror to assign to such a party an arbitrary proportion of the total liability. The introduction of such arbitrary or *773speculative element into the deliberative process can only tend to render the ultimate apportionment unreliable. For this additional reason also, then, I consider it ill-advised to instruct jurors to disregard their ultimate judgment on the issues of negligence and proximate cause in apportioning damages. Whether a statute directing such procedure would be supportable we need not decide. In the absence of a clear legislative direction to such effect, however, I would hesitate to add another feat of mental gymnastics to the already formidable task which each juror undertakes in seeking to determine liability in proportion to fault in a comparative negligence case.
In my view, the requirement of consistency in a juror’s votes in the present context is supportive of the entire deliberative process, encouraging reason instead of speculation in the rendering of mutually interdependent special verdicts upon which the ultimate judgment is to be entered. To sustain that essential consistency in a juror’s vote, it should be required that at least nine identical jurors whose votes ultimately apportion liability between two or more parties also agree that the negligence of each of those parties was a proximate cause of the total damages assessed. The absence of that consistency in the instant case, after full opportunity for deliberation was afforded to the jury, warranted the trial court’s declaration of a mistrial, and we should decline to disturb that ruling.
Kaus, J., concurred.