OPINION OF THE COURT
ALDISERT, Circuit Judge.Defendants have appealed from adverse judgments in a diversity case governed by New Jersey substantive law and arising out of a collision of two automobiles which resulted in the death of plaintiff’s decedent, Dr. Benjamin R. Huddell. Although numerous issues command our attention, the central questions presented involve the nature and extent of an automobile manufacturer’s liability in the context of “crashworthiness” or “second collision” litigation, and the proof required to establish that liability. The jury returned a verdict of $2,024,700 in favor of plaintiff against defendant General Motors, the manufacturer of Dr. Huddell’s car, but found no liability on the part of the defendant Levin, the driver of the other car, or defendant S. Klein Department Stores, Levin’s employer. The district court entered judgment notwithstanding the verdict against Levin and S. Klein. We vacate the judgments against all three defendants, and remand for a new trial.
Treating first the central products liability questions raised by the judgment against G.M., we must determine whether the district court erred in submitting the question of G.M.’s liability to the jury and, if it did not, whether it erred either with respect to evidentiary requirements or in the framing of instructions for the jury on that question. We must also consider the correctness of the judgments entered against defendants Levin and S. Klein. And, finally, we must address several issues presented concerning the calculation of damages: whether income tax effects are to be considered; whether prejudgment interest is to be awarded; and whether inflation may be taken into account.
The pertinent facts have been summarized by the district court:
On the early morning of March 24, 1970, a clear, dry day, Dr. Benjamin R. Huddell, a psychiatrist, was en route from his home in Cherry Hill, New Jersey, to the Delaware State Hospital, where he was engaged in psychiatric research. Dr. Huddell was operating a 1970 Chevrolet Nova, manufactured by General Motors and purchased from its dealer approximately four months prior to the accident. Installed as part of its original equipment were head restraints for driver and front-seat passenger, each at a retail cost of $30.00, whose sole purpose was to prevent rearward rotation of the head and neck in the event of a rear-end collision. Evidence was presented that these head restraints were designed in such a manner as to expose the *732rear of the head to a relatively sharp, unyielding metal edge, covered by two inches of soft, foam-like material.
Dr. Huddell’s car ran out of gas on the Delaware Memorial Bridge, connecting the States of New Jersey and Delaware. His car was brought to a full stop in the left-most, southbound lane of traffic; he was seat-belted in the driver’s seat; and the blinker lights on his vehicle were in operation. At approximately 8:30 A.M., the defendant, George Gerson Levin, in the course of his employment for the defendant, S. Klein Department Stores, Inc., and en route to S. Klein’s branch store in Greenbelt, Maryland, drove his Chrysler sedan at a considerable rate of speed directly into the rear of Dr. Huddell’s Nova. Levin’s speed was estimated at fifty miles per hour (Levin) and sixty miles per hour (plaintiff’s expert). Because of the energy-absorbing characteristics of the vehicles and friction with the roadway, the impact resulted in an acceleration of the Huddell automobile to a maximum speed of 31.7 miles per hour. The rear of Dr. Huddell’s head struck the head restraint at a speed of ten miles per hour.
Levin sustained only superficial injuries for which he was examined, treated and discharged from the hospital within an hour after the accident. With the exception of his head, Dr. Huddell also sustained superficial injuries; the autopsy performed by the Chief Medical Examiner of the State of Delaware, within two and a half hours after Huddell’s death, revealed that his neck, skeletal system and internal organs sustained no injury whatsoever. The blow of his head against the head restraint, however, resulted in “extensive fracture” to the occipital region of the skull. Because of a medical phenomenon known as “contrecoup,” by which the brain of a moving head striking a stationary object sustains injury opposite the point of impact, the frontal portions of Dr. Huddell’s brain were extensively damaged, as a result of which he died one day after the accident.
Dr. Huddell had just completed a residency at the Jefferson Medical College in psychiatry and had opened a private office for the practice of his specialty. At the time of his death he was thirty-nine years of age; his wife Josephine was thirty-four; the range of age of his five children was from three to thirteen years.
Suit was instituted in this court, based upon diversity of citizenship, by Mrs. Huddell in her representative capacity, against George Gerson Levin, driver of the rear-ending vehicle, alleging negligence, against Levin’s employer, S. Klein Department Stores, on a respondeat superior theory, and against General Motors Corporation, charging that the head restraint installed in Dr. Huddell’s vehicle was defectively designed, unreasonably dangerous and failed to give him proper protection against a rear-end collision such as heretofore described.
[In response to a series of interrogatories the jury found] in substance, that Levin was negligent and was acting within the scope of his employment for S. Klein, but that his negligence was not a substantial contributing factor or proximate cause of Dr. Huddell’s death; that Dr. Huddell was not contributorily negligent; that Dr. Huddell’s head did strike the head restraint; that the head restraint was defective and unreasonably dangerous and was a substantial contributing factor of Dr. Huddell's death; that General Motors breached its warranty of fitness which breach was a substantial contributing factor of Dr. Huddell’s death; and that the damages sustained were $2,024,-700.00.
395 F.Supp. 64, 68-70 (D.N.J.1975).
I.
This troublesome case, implicating nascent concepts of state tort liability, demonstrates again the impracticality of the federal diversity forum in the twentieth century and underscores the necessity for Congressional action so eloquently sounded by *733the Chief Justice in his Annual Report on the State of the Judiciary.1 We are to apply New Jersey law, yet we are without the specific guidance of viable New Jersey precedents.2 This appeal requires us to predict how the New Jersey Supreme Court would react when presented with novel and difficult questions of tort law. Specifically we are to predict how that court would view the liability of an automobile manufacturer for the design of a head restraint in a case in which it is alleged that fatal injuries were caused by impact against the head restraint received when the decedent’s stopped car was struck from behind by another car travelling at least 50 m. p. h.
Henningsen v. Bloomfield Motors, Inc., 32 N.J. 358, 384, 161 A.2d 69, 84 (1960), the seminal New Jersey case, stated the basic premise: “[U]nder modern marketing conditions, when a manufacturer puts a new automobile in the stream of trade and promotes its purchase by the public, an implied warranty that it is reasonably suitable for use as such accompanies it into the hands of the ultimate purchaser.” Since 1960 the terminology of liability has changed from “implied warranty” to “strict liability”3 but the jural foundation of liability has remained unchanged. New Jersey has recognized that, as between an implied warranty theory and a strict liability theory, “[t]he governing principles are identic”, Jackson v. Muhlenberg Hospital, 96 N.J.Super. 314, 324, 232 A.2d 879, 884 (1967); and New Jersey has approved the more contemporary terminology of strict liability. Schipper v. Levitt & Sons, Inc., 44 N.J. 70, 90, 207 A.2d 314, 325 (1965); Santor v. A and M Karagheusian, Inc., 44 N.J. 52, 66, 207 A.2d 305, 312 (1965). Accordingly, and in conformity with the essence of the litigation thus far, we treat this appeal under the rubric of strict liability in tort.
Scanlon v. General Motors Corp., 65 N.J. 582, 590-91, 326 A.2d 673, 677-78 (1974), summarized the New Jersey law of strict liability:
The doctrine of strict liability in tort is firmly established in the law of this state. See, e. g., Henningsen v. Bloomfield Motors, Inc., 32 N.J. 358, 161 A.2d 69 (1960); Santor v. A & M Karagheusian, Inc., 44 N.J. 52, 207 A.2d 305 (1965); Schipper v. Levitt & Sons, Inc., 44 N.J. 70, 207 A.2d 314 (1965); Cintrone v. Hertz Truck Leasing & Rental Service, 45 N.J. 434, 212 *734A.2d 769 (1965); Rosenau v. City of New Brunswick and Worthington Gamon Meter Co., 51 N.J. 130, 238 A.2d 169 (1968); Realmuio v. Straub Motors, Inc., 65 N.J. 336, 322 A.2d 440 (1974). Under this doctrine manufacturers and retailers are liable for damages if the product left their hands in a defective condition proximately causing the mishap. Thus, a plaintiff in a strict liability case must establish that the product was defective, that the defect arose while in the control of the defendant and that the plaintiff suffered injury thereby. See Jakubowski v. Minnesota Mining & Manufacturing, 42 N.J. 177, 199 A.2d 826 (1964); Newmark v. Gimbel’s Incorporated, 54 N.J. 585, 258 A.2d 697 (1969); Corbin v. Camden Coca-Cola Bottling Co., 60 N.J. 425, 290 A.2d 441 (1972): Prosser, Law of Torts (4th ed. 1971) § 103.
A product is defective if it is not fit for the ordinary purposes for which such articles are sold and used. Santor v. A & M Karagheusian, Inc., supra, 44 N.J. at 66-67, 207 A.2d 305; see Collins v. Uniroyal, Inc., 64 N.J. 260, 271-272, 315 A.2d 16 n. 6 (1974) (dissenting opinion). Establishing this element requires only proof, in a general sense and as understood by a layman, that “something was wrong” with the product. As a rule the mere occurrence of an accident is not sufficient to establish that the product was not fit for ordinary purposes. However, additional circumstantial evidence, such as proof of proper use, handling or operation of the product and the nature of the malfunction, may be enough to satisfy the requirement that something was wrong with it. See, e. g., Henningsen v. Bloomfield Motors, Inc., supra, 32 N.J. at 409, 161 A.2d 69 (steering malfunction); Cintrone v. Hertz Truck Leasing & Rental Service, supra, 45 N.J. at 452, 212 A.2d 769 (brake failure); Sabloff v. Yamaha Motor Co., Ltd., 113 N.J.Super. 279, 286-287, 273 A.2d 606 (App.Div.), aff’d 59 N.J. 365, 283 A.2d 321 (1971) (wheel locked). Further, a defective condition can also be proven by the testimony of an expert who has examined the product or who offers an opinion on the product’s design.
(Footnotes omitted.)
The essence of the strict liability conception is that liability may be imposed upon the manufacturer without the necessity of proving negligence. A claimant need only demonstrate that at the time it was sold the product was defective, and that the defect caused injury. Those elements established, the care or lack of care of the manufacturer is simply irrelevant. This is not to say, however, that legal precepts derived from negligence law are not useful in strict liability cases. The division of responsibility between judge and jury in negligence cases is analogous to the allocation of responsibility required in a strict liability case. In negligence, the preliminary determination of whether a duty exists with respect to the particular claimant is for the court. The court also must formulate for the jury the general standard of conduct to be applied to the defendant, normally, the reasonably prudent man standard. “[I]t becomes imperative before legal liability for conceded damages can be imposed upon a defendant, for the court in the first instance to inquire and determine the character of the duty which the law under the facts imposed on the defendant as the basis of liability; for manifestly it cannot be conceded that the jury from their inner consciousness may evolve in every variety of tort feasance a legal duty as the standard of liability.” Morril v. Morril, 104 N.J.L. 557, 560-61, 142 A. 337, 339 (1928) (Minturn, J.). The jury, however, must particularize the general standard: it must decide what the reasonably prudent man would have done in this particular case. And, having thus set the particular standard, the jury must decide whether that standard was breached, as well as whether that breach caused the damage to the plaintiff.
Similarly, in the matter before us, the court must make a threshold determination whether there exists any legal obligation at all running from G.M. to the plain*735tiff. And, if there is, the court must decide in the first instance whether, under the circumstances, there is sufficient proof on the issues of defectiveness and causation to submit the case to the jury.
II.
We take it as beyond peradventure that an automobile manufacturer today has some legal obligation to design and produce a reasonably crashworthy vehicle. The manufacturer is not required to design against bizarre accidents; the manufacturer is not required to produce an accident-proof vehicle. But the manufacturer is required to take reasonable steps — within the limitations of cost, technology, and marketability — to design and produce a vehicle that will minimize the unavoidable danger. Rephrased in the terminology of strict liability, the manufacturer must consider accidents as among the “intended” uses of its product. Particularly where safety devices are concerned, the incongruity of viewing the matter otherwise is obvious. See Devaney v. Sarno, 125 N.J.Super. 414, 419, 311 A.2d 208, 210 (1973), aff’d, 65 N.J. 235, 323 A.2d 449 (1974), (“there had to be an accident before the effect of the defective [seatbelt] came into play”). These general propositions are not troublesome. However, the question remains whether there was sufficient proof to put this particular case to the jury. G.M. argues that the district court erred in submitting the issue of its liability to the jury, asserting that no jury question was presented either as to the defectiveness of its product, the head restraint, or as to the proximate cause of the death of Dr. Huddell.
With respect to defectiveness, G.M. argues that the circumstances of the accident — particularly the violence of the high speed collision — preclude any reasonable conclusion that the head restraint was defective. G.M. argues that the severity of the accident is a critical factor in determining whether the head restraint was fit for the ordinary purposes for which it was sold and used: “it could not reasonably be argued that a car manufacturer should be held liable because its vehicle collapsed when involved in a head-on collision with a large truck, at high speed.” Dreisonstok v. Volkswagenwerk, supra, 489 F.2d at 1073.
Plaintiff contended at trial that “[t]be head restraint was defective because its relatively sharp edge of unyielding metal allowed for excessive concentration of forces against the rear of the skull,” (Appellee’s Brief at 8), and that “[t]he head restraint was designed much like an airplane wing, with the front ‘ax-like’ portion aimed directly at the rear of the head”. (Ibid, at 6). In its jury instructions the district court summarized the plaintiff’s claim succinctly: “the head restraint . was defectively designed so as to cause Dr. Huddell’s head, as a result of the rear impact, to come in contact with a relatively thin metal plate rather than a flat surface which would have distributed the force of the impact over a larger area of the skull.” (2487a-88a). Plaintiff’s expert witness, Dr. Paul W. Geikas, testified:
My opinion is that the head rest as we see it here ... is defective, in that its design would allow for an excessive concentration of force on the head.
This is a relatively sharp surface, and what we are striving for is a curve of larger radius, in other words, a more gentle curve, to allow deformation. We want this structure to deform, to contour to the head, to distribute the load, and this would concentrate the load and that is where the problem is. So in my opinion this is defective, yes.
(473a-74a).
We must begin our inquiry with the recognition that the New Jersey Supreme Court teaches that “a defective condition can ... be proven by the testimony of an expert who has examined the product or who offers an opinion on the product’s design.” Scanlon v. General Motors Corp., supra, 65 N.J. at 591, 326 A.2d at 678. Without intending undue cynicism, we are aware of the realities of expert testimony, as we are mindful of the limitations of *736opinion evidence.4 Where the issue concerns a product’s design, however, it would seem that expert opinion is the only available method to establish defectiveness, at least where the design is not patently defective. Moreover, the New Jersey cases manifest a policy liberally favoring jury resolution of defectiveness issues, as well as causation issues, in products liability cases. Moraca v. Ford Motor Co., 66 N.J. 454, 332 A.2d 599 (1975); Newmark v. Gimbel’s Inc., 54 N.J. 585, 258 A.2d 697 (1969); Cintrone v. Hertz Truck Leasing, 45 N.J. 434, 212 A.2d 769 (1965). Plaintiff’s theory was that the head restraint was defective because the metal was not sufficiently deformable and the area of potential impact was unnecessarily small. This theory was not unreasonable. We are persuaded that there was sufficient evidence, albeit opinion evidence, in support of the theory to submit the issue of defectiveness to the jury.
G.M. also argues that, as a matter of law, uncontroverted physical evidence demonstrated that the head restraint could not have caused or contributed to Dr. Huddell’s death. G.M. relies on tests conducted by Dr. Kenneth F. Packer who concluded that an impact sufficient to fracture the skull would have deformed the front interior metal of the head restraint. Because Dr. Packer’s testimony was not contradicted and because the head restraint was not, in fact, deformed, G.M. argues that the question whether the head restraint caused or contributed to the death should not have been submitted to the jury. Plaintiff’s reply is that Dr. Packer’s opinion was based on static, not dynamic, testing.
We feel constrained at this point to express the distinct uneasiness that we feel about this aspect of the case. While there was no deformation of the metal of the head restraint that purportedly killed Dr. Huddell, there was a deformation of the metal of the rear window header: the header, made of the same metal as the head restraint, contained a depression 3" in diameter and Vi" deep. This depression conformed in diameter and depth to the depression caused by a dummy human skull in rear end impact tests. Plaintiff’s experts offered no objective data to explain this phenomenon.
Despite our uneasiness, we must agree with the plaintiff that G.M.’s theory that Dr. Huddell’s head did not strike the head restraint was “a factual contention which, at General Motor’s request, was specifically submitted to and answered by the jury.” (Appellee’s Brief at 17). We have no illusions about the true dimensions of “facts” in judicial proceedings.5 At the same time we are acutely conscious of the limited role allocated to us in reviewing determinations of fact. “The Seventh Amendment bars appellate review of facts found by a jury in actions at common law . .” 9 C. Wright & A. Miller, Federal Practice and Procedure § 2571, at 681 *737(1971). Certainly, the credibility of opinion evidence is for the fact-finder. Sartor v. Arkansas Natural Gas Corp., 321 U.S. 620, 67 S.Ct. 724, 88 L.Ed. 267 (1944). Dr. Packer testified as to a highly scientific matter without bringing to court any papers, records or data supporting his opinion. Particularly in light of this somewhat surprising circumstance, we see no reason to depart from the stated New Jersey rule that “[w]here men of reason and fairness may entertain differing views as to the truth of testimony, whether it be uncontradicted, uncontroverted or even undisputed, evidence of such a character is for the jury.” Ferdinand v. Agricultural Insurance Co., 22 N.J. 482, 494, 126 A.2d 323, 329 (1956). On retrial, the district court may consider the possibility of requiring evidence of testing under dynamic conditions concerning the deformation vel non of a head restraint under the alleged circumstances of this accident.
While we experience much disquietude over the evidence presented on the issue of G.M.’s liability, particularly with respect to the cause of death, we are persuaded that under New Jersey law there was sufficient evidence to submit the issue to the jury. Our conclusion under federal law is not different; we believe there was the minimum quantum of evidence sufficient to present the issue to the jury. Denneny v. Siegel, 407 F.2d 433, 439 (3d Cir. 1969). We are not called upon to consider the limitations that the Seventh Amendment and Rule 38 F.R.Civ.P. place upon state lawmaking competency to decide when questions will be submitted to a federal jury in a diversity case. See ibid, at 437-39. Any such limitations would be limitations on the state’s competency to deny jury trial, limitations imposed to vindicate the strong federal policy favoring jury trial. Byrd v. Blue Ridge Rural Electric Cooperative, Inc., 356 U.S. 525, 78 S.Ct. 893, 2 L.Ed.2d 953 (1957). Where as here, state law liberally favors jury resolution of the issue we see no possibility of conflict with federal policy. Cf. Stoner v. New York Life Insurance Co., 311 U.S. 464, 61 S.Ct. 336, 85 L.Ed. 284 (1940). A jury issue, albeit a thin one, was made out.
III.
A.
That there was sufficient evidence to frame jury questions whether the head restraint was defective and whether the defect was a substantial contributing factor or proximate cause of death is, of course, only a threshold determination in the troublesome resolution of this appeal. We turn now to what we consider a serious and fundamental defect in the plaintiff’s presentation of a case seeking to recover against G.M. for enhanced injuries on a crashworthy or second collision theory.
We have set forth, supra, the basic elements of proof in an orthodox strict liability case under New Jersey law: a defective product causing injury. The simplicity of the general formulation belies the difficulty of applying it in the infinite variety of cases that arise. Unlike orthodox products liability or negligence litigation, crashworthy or second collision cases impugning the design of an automobile require a highly refined and almost invariably difficult presentation of proof in three aspects. First, in establishing that the design in question was defective, the plaintiff must offer proof of an alternative, safer design, practicable under the circumstances. This the plaintiff did: plaintiff’s expert Albert Fonda introduced into evidence the head restraints of several other automobile manufacturers which allegedly were better designed; and it was suggested that G.M.’s head restraint would have been safer if it had merely been turned around so that the “ax-like” portion was not “aimed” at the rear of the driver’s head. Second, the plaintiff must offer proof of what injuries, if any, would have resulted had the alternative, safer design been used. This, in our view, was not satisfactorily established. We agree in this regard with Judge Barlow in Yetter v. Rajeski, 364 F.Supp. 105, 109 (D.N.J.1973) that “it is absolutely necessary that the jury be presented with some evi*738dence as to the extent of injuries, if any, which would have been suffered . had the plaintiff’s hypothetical design been installed . . . Third, as a corollary to the second aspect of proof, the plaintiff must offer some method of establishing the extent of enhanced injuries attributable to the defective design. Having failed to establish satisfactorily the second aspect, a plaintiff necessarily must fail with regard to the third.
Albert Fonda testified that the accident would have been “survivable” if the head restraint had been “designed differently.” (650a). Dr. Geikas similarly testified that the accident would have been “survivable” if the head restraint had been designed “for distribution of load, attenuation of force.” (448a). The district court summarized the testimony: “Plaintiff’s experts testified unequivocally that if the head restraint had been designed to take these principles into account, Dr. Huddell would have survived the accident.” 395 F.Supp. at 75. But the record does not indicate the specific meaning of the term “survivable”, and there is no testimony as to the extent of injuries, if any, which would have resulted in a survivable crash. Although Dr. Geikas testified that there was no evidence of significant injury to vital organs from the accident as it happened, this ignored the possibility that injury to those organs might have been more severe if the great forces of the collision had been more widely distributed over the head and body by an alternate head restraint design. It was not established whether the hypothetical victim of the survivable crash would have sustained no injuries, temporary injuries, permanent but insignificant injuries, extensive and permanent injuries, or, possibly, paraplegia or quadriplegia.
We recognize the serious problems of predicting the direction of state law in troublesome cases like this and, particularly, we understand the difficulties facing a federal trial court forced to make such a prediction. Upon reflection, however, we are constrained to disagree with the district court’s prediction in this case that state law would require G.M. to carry the burden of establishing what damage was attributable to the accident itself as opposed to the allegedly defective head restraint. 395 F.Supp. at 77. The crashworthy or second collision theory of liability is a relatively new theory, its contours are not wholly mapped, but one thing, at least, is clear: the automobile manufacturer is liable only for the enhanced injuries attributable to the defective product. This being the essence of the liability, we cannot agree that the burden of proof on that issue can properly be placed on the defendant manufacturer.
Without proof to establish what injuries would have resulted from a non-defective head restraint, the plaintiff could not and did not establish what injuries resulted from the alleged defect in the head restraint. Without such proof, the jury could not have properly have assessed responsibility against G.M. for the death of Dr. Huddell.
It is well, indeed necessary, to emphasize the fundamental point of departure between our approach and that proffered by the concurrence. We do not perceive the analysis of “second collision” or “enhanced injury” cases to track the legal lore surrounding concurrent tortfeasor actions which, in the concurrence’s formulation, “have combined contemporaneously to cause the injuries”. Concurring opinion, post, at 744. (Emphasis added.) “Second collision” cases do not implicate “clearly established double fault” for the same occurrence. Clearly, if the theoretical underpinnings for liability in this case are to be given effect, Levin may be held liable for all injuries, but General Motors may only be held liable for “enhanced injuries”. Analogies to concurrent actions combining to cause a single impact are simply not applicable. Similarly, analogies to chain collisions are not applicable where, as here, one party is sued on a fault theory for the collision and the other party is sued on the theory of strict liability for the “second collision”. A second source of disagreement between us and Judge Rosenn is the latter’s assumption that “[a] failure in apportion*739ment must then needs excuse both wrongdoers.” Ibid at 744. The burden of apportionment applies only to plaintiff’s claim against General Motors. Should plaintiff fail to meet her burden on this claim, the brute fact is that the negligent driver would not escape liability on the same ground. Traditional negligence concepts determine the ease against Levin and the extent of his liability.
Plaintiff suggested at oral argument, but did not emphasize by brief, that G.M. is precluded from seeking to divide responsibility and limit its liability because the damages at issue are damages recoverable under the New Jersey Death By Wrongful Act6 statute for pecuniary loss to the family flowing from the single, indivisible fact of death. We are not persuaded by this argument but it merits some comment. It was plaintiff, not G.M., who introduced divisibility into the litigation by arguing that the accident was survivable but for the defect in the design of the head restraint. Plaintiff cannot have the argument both ways. Plaintiff may not argue that the ultimate fact of death is divisible for purposes of establishing G.M.’s liability and then assert that it is indivisible in order to deny to G.M. the opportunity of limiting damages. Moreover, to consider the corollary of the argument, if the nature of the proceedings precludes dividing the damages flowing from the death, then Levin and S. Klein may be in a position to contend that they are not liable at all because, by plaintiff’s own proof, Levin’s negligence was survivable. Presumably, no wrongful death claim would lie against one who did not cause death. This argument, of course, boggles the mind. Plaintiff’s brief takes the more reasonable position that “the injury for which recovery was sought . would not have occurred but for the individual’s negligence generating the forces necessary for harm to occur.” (Appellee’s Brief at 24-25). We simply do not accept the proposition that suing for wrongful death suffices to convert limited, second collision, enhanced injuries liability into plenary liability for the entire consequences of an accident which the automobile manufacturer played no part in precipitating.
B.
Our conclusion that a new trial is mandated is buttressed by review of the instructions and charge to the jury. We discern several significant errors which resulted in obvious confusion on the part of the jury.
The verdict returned by the jury, as explained by the special interrogatories, indicated that the jury distinguished the cause of the accident from the cause of the death. They answered “no” to the special interrogatory that asked whether the negligence of Levin was a “substantial contributing factor or proximate cause of the death of Dr. Huddell.” They answered “yes” to the interrogatory that asked whether the “head restraint [was] a substantial contributing factor or proximate cause of the death of Dr. Huddell.” The former interrogatory was confusing because, when combined with the. proof that the accident would have been survivable absent the defective head restraint, it led the jury to conclude that Levin’s actions were not legally the cause of the death. Whether or not Levin was negligent may have been a jury question, but whether or not Levin was causally responsible for Dr. Huddell’s death is hardly a matter on which reasonable men could disagree.7 The district judge *740implicitly recognized this fact when he granted judgment n. o. v. against Levin and Klein’s. The latter instruction, relating to the head restraint, was confusing because it asked whether the head restraint caused the death, not whether the defect in the head restraint caused the death. Accepting arguendo plaintiff’s view that Dr. Huddell’s head struck the head restraint, it is again beyond reasonable disagreement that the head restraint caused the death. If his head had struck the dash board, or the window, or the header, then each of those parts could be said to have caused the death. But that is not the issue: the issue is whether there was a defect which caused injury beyond that which would have otherwise occurred. The instructions as given obscured that issue.
The district court instructed the jury that “if there was any substantial possibility that Dr. Huddell would have survived the collision had the head rest not been defective, assuming you find such a defect, General Motors would be liable if such defect was a substantial contributing factor or a proximate cause of Dr. Huddell’s death.” (App. at 2514a). This instruction, although it properly emphasizes the finding of a defect, misstates the plaintiff’s burden of proof. A plaintiff must prove his case by a preponderance of the evidence; proof of a “substantial possibility” of survival does not comport with that affirmative burden. Moreover, this instruction violated the basic jural conception upon which a crash-worthy case is based, viz., that the automobile manufacturer is liable only for the enhanced injuries occasioned by the defectiveness of its product.
Over objection, the court instructed the jury:
I charge you, however, that the relative severity of the impact is relevant only to your consideration of proximate cause, not to your consideration of whether there was a defect in the design of the head rest. (App. at 2513a).
This instruction, we believe, was error. If a products liability case is submitted to the jury, it is the jury who must determine whether the product was “reasonably fit for the ordinary purposes for which such articles are sold and used.” Santor v. A and M Karagheusian, Inc., supra, 44 N.J. at 67, 207 A.2d at 313. The “relative severity of the impact” goes to the heart of the issue of defectiveness in terms of the “ordinary purposes” for which the product, the head restraint, was designed.
We emphasize here the unique aspect of “defectiveness” presented by a crashworthy case. Ordinarily, where a product malfunctions and itself precipitates injury, there is little problem in establishing a defect; the issue frequently is whether the defect is traceable to the defendant manufacturer, or whether the defect arose after the product left the manufacturer’s control, e. g., by long usage of the product. See Moraca v. Ford Motor Co., 66 N.J. 454, 332 A.2d 599 (1975). However, in a crashworthy case, impugning the design of the product in question, the difficulties are reversed. There can be no doubt that the design of the product is traceable to the manufacturer. The central issue is: was the product “defective”? And this can only be evaluated in the context of a particular risk. A manufacturer is required to design a product “reasonably” fit for the “ordinary” purposes for which it is sold and used. In the context of safety equipment, this means that the manufacturer is not required to design against extraordinary accidents of unusual circumstance or severity. On this point, we question the relevance of plaintiff’s assertion that “[i]f a seat belt is faulty, it remains faulty whether an accident occurs at 5 m. p. h. or 100 m. p. h.” *741(Appellee’s Brief at 21). This may be true, but it begs the question: the question is whether or not the seat belt or the head restraint is “faulty”. If the seat belt did not adequately protect its wearer in a 5 m. p. h. crash, then a proper inference might be drawn that the belt was defective; but if a man were killed wearing his seat belt in a 100 m. p. h. crash, could it be argued with the same assurance that the belt was defective? At least in the context of safety design, we see no meaningful way to evaluate the defectiveness vel non of a product except in the context of a particular risk. For that reason, we believe that the District Court’s instruction on this point constituted error.
Because of the serious and fundamental deficiency in plaintiff’s proof, and because of the several errors in the jury instructions, we believe that a new trial is required as to the existence, and the extent, of liability on the part of G.M.
IV.
We now address the entry of judgment notwithstanding the verdict in favor of the plaintiff and against Levin and his employer, S. Klein. The jury found that Levin was negligent and that he was acting in the scope of his employment for S. Klein, but that his negligence was not the proximate cause of Dr. Huddell’s death. Subsequently, and without benefit of a motion for a directed verdict as required by the letter of Rule 50, F.R.Civ.P., the district court entered judgment notwithstanding the verdict against Levin and S. Klein.
Under New Jersey practice the test for taking an issue from the jury has been stated as follows:
[W]hen the testimony of witnesses, interested in the event or otherwise, is clear and convincing, not incredible in the light of general knowledge and common experience, not extraordinary, not contradicted in any way by witnesses or circumstances, and so plain and complete that disbelief of the story would not reasonably arise in the rational processes of an ordinarily intelligent mind, then a question has been presented for the court to decide and not the jury.
Ferdinand v. Agricultural Insurance Co., supra, 22 N.J. at 494, 126 A.2d at 330. The proper reluctance of courts to decide negligence issues was not involved here because the jury found that Levin was negligent; judgment notwithstanding the verdict was granted because of the district court’s view of the causation issue. Moreover, we agree with the district court’s view, implicit in its action, that having found Levin negligent the jury could not properly conclude that his acts were not a proximate cause of the death. However, as we have indicated, we believe that the jury’s strange conclusion resulted, at least in part, from confusing and incorrect instructions as to the law. While we might take a different view of the matter if the erroneous judgment against General Motors were not implicated, we are convinced that it would be improper under these circumstances to sustain the judgment against Levin and S. Klein while vacating the judgment against G.M. The issues were inextricably intertwined. We are not confident that the confusion apparent in the jury’s answers to certain of the special interrogatories did not permeate its consideration of the entire case. We believe, therefore, that a new trial is required as to Levin and S. Klein as well as G.M.
A seriously contested issue before us is whether the failure of a condition precedent, to-wit, a motion for a directed verdict as set forth in Rule 50(a), F.R.Civ.P., prevented the district court from entering judgment notwithstanding the verdict under Rule 50(b) against Levin and S. Klein. Plaintiff and General Motors contend that plaintiff’s request for binding instructions was the equivalent of a motion for a directed verdict and as such provided the procedural predicate for the Rule 50(b) judgment. Because we have determined that the judgment entered against Levin and S. Klein must be vacated on other grounds, we need not and do not here decide the issue.
Upon retrial, the district court may request the parties to consider whether *742the New Jersey Supreme Court would be receptive to a rule kindred to the apportionment rule announced by the New York Court of Appeals in Dole v. Dow Chemical Co., 30 N.Y.2d 143, 331 N.Y.S.2d 382, 282 N.E.2d 288 (1972) that “where a third party is found to have been responsible for a part, but not all, of the negligence for which a defendant is cast in damages, the responsibility for that part is recoverable by the prime defendant against the third party. To reach that end there must necessarily be an apportionment of responsibility in negligence between those parties. The adjudication is one of fact and may be sought in a separate action . . . or as a separate and distinguishable issue by bringing in the third party in the prime action.” See Leflar, Contribution and Indemnity Between Tortfeasors, 80 U.Pa.L.Rev. 130 (1932). The Dole decision itself involved a defendant-third party relationship, but the apportionment principle was subsequently applied to cross claims among co-defendants in Kelly v. Long Island Lighting Co., 31 N.Y.2d 25, 334 N.Y.S.2d 851, 286 N.E.2d 241 (1972). Dole implicated active and passive negligence, and abolished the distinction. It did not implicate a combination of negligence and products liability; and it did not implicate the troublesome — and in our view sui generis — concept of second collision liability. Dole did not affect the plaintiff’s rights at all; it only provided for an apportionment of responsibility among defendants. But it did represent a salutary judicial reevaluation of a tired common law doctrine that had long since outlived its purposes. The common law must accommodate changing conditions, new rights and remedies. New theories of liability, like the second collision theory, require fresh legal thinking on the part of courts. Whether old precepts ought to be retained, or discarded, in second collision cases is a matter that must abide the fermentation of the law in the state court systems. Meanwhile, we of the federal system are placed in the unhappy position of making predictions, and merely predictions, in these diversity cases, without authority to our conclusions and without the jural responsibilities normally vested in a court system.
V.
We turn now to a number of issues presented in this appeal relating to the calculation of damages for wrongful death. G.M. contends that the district court erred in denying defendants the opportunity to present evidence of the decedent’s income tax expenses. Gross earnings evidence alone was permitted to be introduced. Not having the benefit of a definite ruling of the state’s highest court at the time of trial, the district court was “persuaded that the New Jersey Supreme Court, if directly confronted with the issue, would continue to adhere to the gross earnings rule . ..” 395 F.Supp. at 87. On June 18, 1975, the court decided the issue in Tenore v. Nu Car Carriers, Inc., 67 N.J. 466, 494-95, 341 A.2d 613, 628-29 (1975):
Consequently, we hold that under our wrongful death act, defendants must have an opportunity to cross-examine plaintiffs’ witnesses to elicit testimony concerning deceased’s income tax liability, or to develop the matter by extrinsic evidence, to the end that the jury be enabled to make an informed estimate, based upon the deceased’s projected net income after taxes, of the survivor’s pecuniary loss. Consequently, plaintiff’s recovery must be calculated on the basis of the deceased’s net income after taxes giving due regard to the evidence adduced on the deceased’s income tax liability.
(Footnotes omitted.)
Inasmuch as a new trial is being ordered for other reasons, we need not meet the question of Tenore’s retroactivity or prospectivity. Since this issue is controlled by New Jersey substantive law, if the question of damages is reached at retrial the district court will be bound to apply the rule of Tenore.
G.M. further contends that the district court erred in its construction of New Jersey law with respect to imposing prejudgment interest. We find no error in the district court’s disposition of this point and *743agree with its reasoning set forth at 395 F.Supp. at 93-94, permitting imposition of prejudgment interest at a rate of 8 per cent per annum in accordance with New Jersey Rule 4:42-ll(b) as amended April 1, 1975.
Finally, G.M. argues that the district court erred in admitting evidence that the inflationary trend of the general economy would cause Dr. Huddell’s earnings to increase by 6 per cent per annum. We have had occasion to consider this issue in other contexts, Griffin v. United States, 500 F.2d 1059, 1070 (3d Cir. 1974); Hoffman v. Sterling Drug, Inc., 485 F.2d 132, 143-44 (3d Cir. 1973); Frankel v. Heym, 466 F.2d 1226, 1229 (3d Cir. 1972), but this is our first occasion to consider the issue in the context of New Jersey law. Again, Tenore v. Nu Car Carriers, Inc., supra, 67 N.J. at 481, 341 A.2d at 621, controls: “in wrongful death actions, the jury should be instructed that it is permissible for them to consider the impact of inflation upon the survivor’s future losses.” Expressing concern that expert evidence on this question might be “unduly speculative”, the Court set forth guidelines for the reception of such evidence:
To reduce the possibility of undue conjecture on the part of the jury, plaintiff should also have an opportunity to offer expert economic testimony on the question to provide them with informed guidelines in their deliberations. See Evid.R. 56; Rempfer v. Deerfield Packing Corp., 4 N.J. 135, 141-42, 72 A.2d 204 (1950). Of course, we recognize that economists may differ in their analyses of past trends and future projections and consequently we anticipate that defendants in wrongful death cases will want to offer the testimony of their own experts in rebuttal. This is a familiar practice in other contexts and we see no reason why the same procedure cannot be resorted to in order to give the jury in death cases the benefit of all available points of view.
In holding that expert economic testimony is admissible in actions for wrongful death on the issue of the effect of inflation on the survivor’s future pecuniary loss, we are aware that not all courts share this view. In our judgment, however, it is simply unrealistic to ignore the problem of inflation in this context. Given the enormity of the potential impact of rising costs over a long period of time, we conclude that evidence of informed estimates of future inflationary trends will result in damage awards that are fair and just.
We return to the matter of introduction in evidence of tables prepared by the expert witness purporting to show plaintiff’s aggregate damages. This is improper for two reasons. In the first place, the tables assume fact findings not within the peculiar expertise of the economic expert. In the second place, the projection of a gross figure before the jury submitted by an expert tends to exert an undue psychological impact leading to the danger of its uncritical acceptance by the jury in the place of its own function in evaluating the proofs.
What we do expect is that the experts will provide the jury with their analyses of trends of future wage increases and discount interest rates generally and that then, giving due regard to credibility, the jury will use those trends and rates in arriving at their own independent single-figure appraisal of plaintiff’s pecuniary loss.
Ibid., 67 N.J. at 481-84, 341 A.2d at 621-23 (footnotes omitted).
Upon retrial the Tenore inflation rule and guidelines will apply.
Our reference to New Jersey law concerning the different elements involved in calculating the damage award must not be in derogation of the established principle in this federal circuit that a jury award may not be based merely on speculation or conjecture. Russell v. City of Wildwood, 428 F.2d 1176 (3d Cir. 1970). We are mindful of the difficulties of calculating damages in a case like this; mathematical exactitude is not required. The plaintiff, however, bears the burden of proof and it is the responsibility of the plaintiff to provide *744for the jury some evidentiary and logical basis for calculating or, at least, rationally estimating a compensatory award. “[SJheer conjecture cannot be the basis of a jury finding.” Dixon v. Pennsylvania R. R., 378 F.2d 392, 394 (3d Cir. 1967).
We have considered all the contentions raised by the parties. The judgment of the district court will be vacated as to all defendants and the proceedings remanded for a trial de novo8 in accordance with the foregoing.
. The Congress has not yet acted on the matter of federal court jurisdiction in diversity of citizenship cases although numerous responsible studies, including that of the American Law Institute, which has now been available for seven years, called for change in this outmoded procedure. I have an obligation to repeat what I have said before, that with a few exceptions which can be dealt with as such, diversity cases have no more place in the federal courts in the second half of the 20th century, and surely not in the final quarter of this century, than overtime parking tickets or speeding on the highways simply because the street or highway is federally financed.
Annual Report on the State of the Judiciary, presented to the American Bar Association, Philadelphia, Pa., February 16, 1976 (summarized at 44 U.S.L.W. 2389) (footnote omitted).
. The major cases involving the second collision theory of liability have arisen in federal courts, often in the context of negligence; these cases furnish, at best, unauthoritative and diverse prognostications as to how several state high courts would rule under the circumstances. Wooten v. White Trucks, 514 F.2d 634 (5th Cir. 1975) (Kentucky); Dreisonstok v. Volkswagenwerk, 489 F.2d 1066 (4th Cir. 1974) (Virginia); Larsen v. General Motors Corp., 391 F.2d 495 (8th Cir. 1968) (Michigan); Evans v. General Motors Corp., 359 F.2d 822 (7th Cir. 1966) (Indiana); Hardy v. Volkswagen of America, 65 F.R.D. 359 (W.D.Pa.1975) (Pennsylvania); Yetter v. Rajeski, 364 F.Supp. 105 (D.N.J.1973) (New Jersey); Dyson v. General Motors Corp., 298 F.Supp. 1064 (E.D.Pa.1969) (Pennsylvania).
. The strict liability doctrine imposes in effect a warranty obligation upon the manufacturer, but is intended to remove contractual impediments, e. g., privity, that have become associated with warranty actions. “There is nothing in [Section 402A] which would prevent any court from treating the rule stated as a matter of ‘warranty’ to the user or consumer. But if this is done, it should be recognized and understood that the ‘warranty’ is a very different kind of warranty from those usually found in the sale of goods, and that it is not subject to the various contract rules which have grown up to surround such sales.” Restatement (Second) of Torts § 402A, comment m (1965).
. Opinion evidence has long been skeptically regarded as a dubious kind of proof. “It is mere opinion, which is not evidence.” Carter v. Boehm, 3 Burr. 1905, 1918 (1766) (Mansfield, J.). Courts today may be less dogmatic, but the skepticism remains. See generally McCormick on Evidence §§ 11, 12 (1972). “Of all forms of evidence, opinion evidence is the weakest and least reliable. Even though uncontradicted it need not be accepted.” Blue Mountain Tel. & Tel. Co. v. Pennsylvania Public Utility Comm’n, 165 P.Super. 320, 327, 67 A.2d 441, 444 (1949). Also see Minnesota Mining and Manufacturing Co. v. Berwick Industries, Inc., 532 F.2d 330, 333 (3d Cir. 1976).
. The problem is that the facts are forever gone and no scientific method of inquiry can ever be devised to produce facsimiles that bring the past to life. The judicial process deals with probabilities, not facts, and we must therefore be on guard against making fact skepticism our main preoccupation. However skillfully, however sensitively we arrange a reproduction of the past, the arrangement is still that of the theater. We acknowledge as much when we speak of re-enacting the crime or the accident or perhaps some everyday event; we know better than to speak of reliving it. The most we can hope for is that witnesses will be honest and reasonably accurate in their perception and recollection, that triers of fact will be honest and intelligent in their reasoning, and that appellate courts will frame opinions with enough perspective to guide others in comparable fact situations and preclude their disputes from festering into litigation.
Roger J. Traynor, Fact Skepticism and the Judicial Process, 106 U.Pa.L.Rev. 635, 636 (1958).
. N.J.S.A. 2A:31-1. Plaintiff’s complaint alleged in several parts that Dr. Huddell “was made to suffer great physical pain before his death”, apparently as a predicate to recovery in Dr. Huddell’s own right under the New Jersey Survival statute, N.J.S.A. 2A:15-3. However, the importance of the survival aspect of the case was disclaimed by plaintiff’s counsel at oral argument and it is apparent that the significant damages in contention on this appeal are the wrongful death damages, damages measured by “the pecuniary injuries resulting from such death to the persons entitled to any intestate personal property of the decedent.” N.J. S.A. 2A:31-5.
. With minor variations, the consensus of testimony was that the speed of the Levin Chrysler at impact was 50 to 68 m. p. h., that Dr. Huddell’s Nova accelerated almost instantly from 0 to 32 m. p. h., that the collision generat*740ed 1100 to 1500 pounds of force on the occipital area of the skull, and that Dr. Huddell’s Nova was compacted by 4' 10" as a result of the collision. Under these circumstances it is simply offensive to the rational faculty to say that Levin’s negligence — assuming he was negligent — was not a proximate cause of Dr. Huddell’s death. Cf. Pappas v. Santiago, 66 N.J. 140, 143, 329 A.2d 337, 339 (1974). Common sense is not out of place in the law. Davis v. Washington, 168 U.S.App.D.C. 42, 512 F.2d 956, 966 (Robb, J., dissenting) cert. granted, 423 U.S. 820, 96 S.Ct. 33, 46 L.Ed.2d 37, 44 U.S.L.W. 3200 (Oct. 6, 1975) (No. 74-1492).
. We do not disturb the jury’s findings that Dr. Huddell was not contributorily negligent and that Levin was acting within the scope of his employment for S. Klein (at 732 supra). Upon retrial those two findings will not be open to relitigation.