dissenting.
Recognizing that we are dealing with issues that have come to the fore after a lengthy trial and subsequent to the painstaking efforts of the trial judge to resolve difficult questions about the development of tort law in North Carolina, it is with some hesitance that I dissent from the panel’s disposition of this appeal.
I.
At a minimum, I believe that removal of the strict liability claim from the controversy necessitates a remand for a new trial on damages. As the majority correctly notes, the Erie principle mandating that state substantive law governs in diversity actions, requires that “until such time as a case is no longer sub judice, the duty rests upon federal courts to apply state law ... in accordance with the then controlling decision of the highest state court.” Vandenbark v. Owens-Illinois Co., 311 U.S. 538, 543, 61 S.Ct. 347, 350, 85 L.Ed. 327 (1941) (emphasis added); Baker v. Outboard Marine Corp., 595 F.2d 176 (3d Cir. 1979). Consequently, the recent state supreme court decision in' Smith v. Fiber Control Corp., 300 N.C. 669, 268 S.E.2d 504 (1980), handed down subsequent to the trial court judgment, which refused to adopt the principle of strict liability in tort, vitiates the district court’s prediction that North Carolina would embrace the strict liability theory. I therefore agree with the majority’s holding that it was error for the district court to permit the jury to consider the plaintiffs’ strict liability claim. But I cannot adhere to the ultimate conclusion reached by the majority that the misplaced reliance on the strict liability theory in no way affected the outcome of the litigation. For assuming arguendo that VW is still liable on the crash-worthiness theory, it does not follow that the damages assessed against VW would remain unaltered after the strict liability claim has been eliminated.
Two alternate denominations for the crash-worthiness concept — the enhancement or second collision theories — further emphasize that liability under such a doctrine does not make one responsible for all injuries occasioned by the initial accident, but only for those that could have been prevented by an alternate, safer design. See Huddell v. Levin, 537 F.2d 726, 740 (3d Cir. 1976); Dreisonstok v. Volkswagenwerk, A.G., 489 F.2d 1066, 1069, 1076 (4th Cir. 1974).
As the trial judge explained to the jury, VW was potentially liable under two tort theories, but the damages recoverable under each claim involved different amounts. Thus, as the judge correctly instructed on the strict liability count, “if you find there was a defect in the right rear ball cage at the time of its manufacture and sale, and that this defect was the proximate cause of the accident, you are to award damages to the plaintiffs in an amount that will compensate them for all the injuries they sustained as a result of the entire accident.” (emphasis added) In contrast to the strict liability theory, under which VW would be responsible for all ensuing damages, under the crashworthiness theory VW would be liable only for injuries which exceeded those that would have occurred had the manufacturer designed a reasonably safe and suitable vehicle. See Huddell v. Levin, 537 F.2d 726, 740 (3d Cir. 1976).1 That is, in the words of the trial judge,
you are to award damages under the second theory (if you only find under the second theory) only for the enhanced injuries which plaintiffs received as a result of any negligent design which you find. You are not to award plaintiffs damages for the total amount of injury they sustained as a result of the entire accident if you find the defendant liable only on the second theory of liability.” Trial Transcript 7764. (emphasis and parentheses added)
Having excised the strict liability claim from the case, I do not understand how the majority can now ignore both case law and the plain directive of these instructions, which make clear that VW is not liable for damages in full when crashworthiness is the only foundation for negligence.
*851It is suggested that defendants are foreclosed from raising this damage point because they did not object to the instructions or interrogatories. The defendants however, cannot be faulted for failing to object to the instructions dealing with liability. Their position throughout the trial, as well as on appeal, was that North Carolina law did not embrace strict liability or crashworthiness doctrines, and that the trial court erred in its analysis of the probable evolution of state law. The problem we address here arose after the instructions had carefully distinguished the two causes of action; it comes about primarily as a result of the interrogatories that were posed to the jury by the trial judge. Once the jury found VW liable under each distinct theory of liability, it was then simply asked to assess damages with respect to each individual plaintiff in a lump sum. Notwithstanding the jury finding of liability under both the strict liability and crashworthiness theories, on a hindsight basis it may be contended that it would have been preferable to insist on two separate damage assessments given that the instructions on liability and proximate cause had addressed each claim individually.2 But to penalize VW for not demanding two distinct computations, and to require VW to pay for damages not authorized by North Carolina law, would place an unfair burden on defendants in the unusual situation here.
Inasmuch as the North Carolina Supreme Court had not settled at the time of trial either the availability of strict liability or of crashworthiness causes of action in North Carolina, VW had consistently attacked the district court’s predictions respecting the eventual acceptance of each of these doctrines. And on appeal VW continued to argue that erroneous legal predictions and instructions with respect to either theory so infected the trial that judgment for defendants, or a new trial, was required. Conversely, plaintiffs maintained that if the verdict is sustained as to either of their theories, then the judgment in toto should be affirmed.3 Neither party contemplated anything other than an all-or-nothing judgment. In holding VW liable for damages in full, then, based on its failure to demand separate computations in the form of discrete interrogatories, the majority is in effect requiring the defendants to foresee the nearly inscrutable: namely, that the appellate court would overturn the district court’s legal forecasts regarding North Carolina law with respect to strict liability but not concerning crashworthiness. Any other outcome — upholding both theories, upholding the strict liability but not the crashworthiness, or invalidating both theories— would have produced the all-or-nothing results that the respective parties envisaged.
Moreover, it was the plaintiffs who proceeded under two theories of liability — a course ultimately proved by the highest court of North Carolina to be incorrect. And as prescribed by Huddell, under the crashworthiness theory the burden was on the plaintiffs to establish what injuries were attributable to the accident itself and what injuries were sustained solely on account of the defective window system. See Huddell v. Levin, 537 F.2d 726, 738 (3d Cir. 1976). Because, under Huddell, the applicable procedure required plaintiffs to prove enhanced injuries — that is, those that would not have occurred but for the defective design — it would appear reasonable that plaintiffs be required to carry the parallel burden of protecting their liability theory with sufficiently specific interrogatories on damages. It would seem unfair, at least to me, to permit plaintiffs’ pursuit of two causes, one valid and one invalid, to convert the defendants’ limited, second collision liability into plenary liability for all consequences of the accident. Yet, with the *852present disposition, the majority does just that.4
In the absence of a remand, the Court is faced with either attempting to segregate two damage figures from an indivisible lump sum, a task beyond our ability to perform, or with making the implausible assumption that all the damages incurred by the plaintiffs resulted from the defective window retention system. Yet this hypothesis rests on the belief that had VW designed a reasonably safe and suitable window system none of the injuries for which the jury assessed damages would have occurred. The majority, pointing to testimony by one expert witness that had the windows not been negligently designed no injuries whatsoever would have been incurred, apparently adheres to this supposition. See maj. op. infra at 849. However, this ignores the testimony of another expert, who was unwilling to speculate as to whether the injuries received by some of the plaintiffs occurred inside or outside the van.5 If the injuries were suffered inside the van, they would have occurred regardless of the allegedly defective window system. Moreover, the majority’s hypothesis assumes that the jury drew the same conclusion as the one expert, an inference which, as I view it, is impermissible for an appellate court to make.
In attempting to buttress its position, the majority notes that Alexander Maldonado, the one passenger who was not ejected, suffered only minor scratches and bruises. However, the evidence suggests that Maldonado escaped injuries by falling into the space beneath the dashboard, an area capable of protecting only one person.6 Not only is it uncertain that a reasonably safe, alternative window design would have prevented the ejection of the plaintiffs — in fact, there was considerable evidence to the contrary7 — but it is far from clear that plaintiffs’ fate, had they managed to stay inside the vehicle, would have resembled that of Maldonado.
While the rationale underlying the crash-worthiness theory — that manufacturers should be liable for accidents regardless whether the defect actually caused or merely enhanced the damages — has merit,8 such a theory was not intended to impose upon manufacturers the duty to produce a vehicle incapable of exacerbating injuries.9 Rather, liability under the crashworthiness theory extends only to the injuries that a safer, economically feasible design reasonably would have prevented — that is, had the window retention system been somewhat sturdier and more resilient. Cf. Dawson v. Chrysler Corp. 630 F.2d 950 (3d Cir. 1980). Such a theory of liability, as articulated in Huddell, requires a remand for an assessment of damages for injuries suffered solely on account of the negligent design in the window retention system.
II.
The second aspect of the majority opinion which I find disquieting is the analysis of the crashworthiness claim that the trial court originally adopted and that the majority accepts today. For the district court, in removing all seat belt considerations from the case,10 and the majority here, in agreeing that “it would have been irrelevant for the jury to consider the use of seat belts in deciding whether the window retention system was negligently designed,” ap*853pear to have misconstrued the concept of crashworthiness. Simply acknowledging that the jury found the negligently designed window retention system to have proximately caused the plaintiffs’ injuries, as the majority seems to do, does not relieve us from the obligation of posing the prior inquiry: was it error for the trial judge to eliminate totally the presence of seat belts from the jury’s considerations of the crash-worthiness of the van?
I fully agree that North Carolina law, as set forth in Miller v. Miller, 273 N.C. 228, 160 S.E.2d 65, 70 (1968), precluded in this type case a contributory negligence defense with respect to the initial accident based on a failure to wear seat belts. But Miller in no way prohibits the introduction of evidence concerning the presence of seat belts insofar as such evidence pertains to the reasonableness of the van design for purposes of the crashworthiness or second collision theory. Indeed, the district judge attempted to adhere to this view in his early instructions to the jury. He explained:
To the extent that seat belts are relevant at all in this case, their relevance is only as to whether or not it was reasonable for those who designed the retention power of these windows to assume, if they did assume, that occupants of Yolkswagens, type 2 van, would wear seat belts.... In other words, in deciding whether or not there was negligence in the design of the window, you have a right to take into consideration whether or not in designing the window the architects or the designers took into account that people would be wearing seat belts and perhaps even more importantly for this case, whether or not, if they did assume people would be wearing seat belts, such an assumption was reasonable on their part. Trial transcript 2103-04.
Even this perspective — which focuses on the rationality of VW’s assumption that people might wear seat belts — may be an unduly narrow view of the crashworthiness concept.
For the question of defectiveness for purposes of a crashworthiness analysis is a uniquely contextual one. As Judge Aldisert stated for the Court in Huddell v. Levin: '
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.... 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.
537 F.2d at 740.
Thus, for the jury to determine whether the window system was reasonably designed for the purposes for which it was sold requires expanding the scope of inquiry beyond the abstract seatbelt-window design relationship. An evaluation of the possible defectiveness of the window system, in light of the vehicle as a whole and in the particular circumstances of a rollover, is necessary.
Similar to the test advanced by Judge Aldisert in Huddell, other courts have also emphasized that the jury be instructed in resolving defective design issues to consider the vehicle as a whole in the context of a given accident. Thus, in Dreisonstok v. Volkswagen, A.G., supra, in which plaintiffs sued for enhanced injuries sustained when a VW microbus crashed into a telephone pole, the court noted that any crashworthy case necessarily involves a balancing of many factors — including price, particular type of vehicle, and circumstances of the accident— “to determine whether the manufacturer had used ordinary care in designing a car, which, giving consideration to the market purposes and utility of the vehicle, did not involve unreasonable risk of injury to occupants ...” 489 F.2d at 1073. And as the court in Dyson v. General Motors Corp. recognized, the differentiation between various models means “that a convertible can not be made as safe in roll-over accidents as a standard four-door sedan with center *854posts and full-door frames.” 298 F.Supp. 1064, 1073 (E.D.Pa.1969). To the same effect, in Wilson v. Volkswagen of America, 445 F.Supp. 1368 (E.D.Va.1978), which involved a single car rollover, the court viewed crashworthiness as a holistic, contextual concept: “any vehicle should be made as safe as it reasonably can, considering its special purpose and ‘intended use’; but standards of safety themselves must take into account the utility of the vehicle.” 445 F.Supp. at 1372. Crashworthiness then relates back to the reasonableness of the entire vehicle at the time of its manufacture, although it is necessarily assessed in the context of an actual accident, which may not have been foreseeable.11
With respect to the crashworthiness claim before us, the existence of seat belts was relevant to the question of reasonableness of the window design in the context of a rollover accident. For this reason, the ruling by the district court that totally removed such evidence from the jury, constitutes error.12 The majority opinion has excerpted a portion of the pointed interchange between the district court and VW’s expert witness, Mr. Pohl, in which Mr. Pohl stated that the window retention system was designed without regard to whether the occupants would be wearing seat belts. See maj. op. supra at 843-844. This colloquy formed the basis for the court’s conclusion that, inasmuch as yolkswagen, in designing its windshield and window retention system, did not assume that occupants would be wearing seat belts, evidence relating to the availability of seat belts was improper. See Trial Transcript at 7759. Given the Huddell directive and the above discussion of crashworthiness, however, this assumption does not dispose of the question whether in the context of this particular risk — a rollover at a high speed — the window design plus the existence of the seat belts provided a reasonably crashproof vehicle.
While Mr. Pohl’s response is arguably subject to several interpretations, no possible construction justifies removing all seat belt considerations from the jury’s determination of crashworthiness. The statement by Mr. Pohl plausibly suggested that VW neither assumed that passengers would nor assumed they would not wear seat belts. It says nothing about whether in a high-speed rollover situation seatbelts are perhaps the most reasonable and efficient means of protecting against injuries, and whether any window design can guard against accidents of such severity. In fact, examination of the entire course of Pohl’s testimony manifests a position that it was unreasonable to rely upon the windows to retain occupants in a rollover accident.13 It would therefore, at least to me, appear improper on the crashworthiness issue to remove evidence regarding the availability of seatbelts and the reasonableness of plaintiffs’ nonuse of them from the jury based on the isolated statement of an expert witness that the VW window retention system was designed without regard to whether occupants would or would not be wearing seat belts.
III.
As the entire concept of crashworthiness indicates, each vehicle constitutes a special design and each accident constitutes a rath*855er singular circumstance, often posing problems not easily analogized to earlier torts cases. The consequent difficulty in providing comprehensive standards to govern such a range of contingencies coupled with continuing technological advances and design modifications, may explain why Congress did not intend federal motor vehicle safety standards to be all-inclusive. As we acknowledged in Dawson v. Chrysler Corp., supra, Congress explicitly provided that “ ‘Compliance with any Federal motor vehicle safety standard issued under this sub-chapter [National Traffic and Motor Vehicle Safety Act] does not exempt any person from any liability under common law.’ 15 U.S.C. § 1397(c) (1976).” 630 F.2d 950, 958 (3d Cir. 1980). Yet the problems generated by the present ease ultimately serve to underscore the perception in Dawson: the existing system for assessing liability and compensating injuries in automobile accident cases, which permits individual juries under different state systems to arrive at discrepant outcomes that impose conflicting standards on manufacturers, is far from efficient or fair.
The case before us illustrates the inefficiencies and inequities created by the uneven evolution of state tort law doctrines and the absence of uniform federal standards in the automobile field. The conscientious efforts of the district court here to anticipate the course of North Carolina tort law proved futile. A state supreme court decision handed down after the trial was completed peremptorily negated a careful attempt at judicial extrapolation. Divergent state tort laws produce similarly arbitrary, stochastic outcomes for injured plaintiffs. Had the accident occurred in any state but North Carolina, and two others which have not yet adopted section 402A of the Restatement, a strict liability cause of action would have been available. And predictability and uniformity in the governing law, quite obviously, would have eliminated many of the difficulties which confront us on this appeal.
As we indicated in Dawson, the irrationalities in the existing system affect the automobile manufacturers as well. The ultimate effect of jury verdicts which reach beyond federal safety standards and are arrived at under different state laws, is the imposition of conflicting requirements on producers attempting to satisfy a nationwide market. In the course of the present trial VW found itself caught between Scylla and Charybdis, so to speak. On the one hand, VW’s obligations were measured by national statistics which are arguably inappropriate for this type accident or this type van.14 On the other hand, the jury assessing liability was guided by insufficiently articulated state law. Yet, VW is now bound by this jury verdict, arrived at under North Carolina law and based on potentially inapplicable federal standards, a verdict which may prove to be at variance with what some later jury in another state decides. The absence of precise and comprehensive national standards, exacerbated by the vagaries of different states' laws, leaves us far from the goal of reducing and fairly distributing the sum of accident costs as well as the costs of reducing the number of future collisions.
It would appear, then, that injured parties, manufacturers and the public would all benefit from greater certainty and consistency in the governing law and the concommitant efficiency in its administration. Indeed, manufacturers might well consent to more rigorous safety standards in exchange for a guarantee that they would be applied nationwide in an evenhanded manner. And equally important, the public would benefit from a concerted attempt to reach a national consensus on the conflicting goals of productivity, energy conservation and passenger safety.
*856Unfortunately, given the existing adjudicative system, I am compelled to comment primarily upon the appropriateness of a remand for a reassessment of damages, and to point out why it appears necessary to retry the crashworthiness claim to permit incorporation of seat belt evidence into the jury assessment of the reasonableness of the vehicle’s design.
. The trial court and counsel for plaintiffs agreed that the burden of proof with regard to the crashworthiness claim would be guided by Huddell v. Levin, 537 F.2d 726 (3d Cir. 1976). See J.A. 490.
. Fed.R.Civ.P. 51, which precludes a party from assigning “as error the giving or failure to give an instruction unless he objects thereto,” refers to instructions but makes no mention of interrogatories.
. Although this may represent an aggressive plaintiff strategy, it ignores significant differences in the damages recoverable under the two liability theories.
. As to the majority’s contention that VW did not raise on appeal the point that damages under strict liability and crashworthiness theories amount to two distinctly different sums, it is suggested that VW’s argument that the instructions on strict liability and crashworthiness were erroneously given and infected the entire proceedings with prejudicial error is sufficiently broad to encompass this claim. See Defendants-Appellants Brief at 11, 12 n.2, 14.
. See J.A. 1125-26.
. See Defendants-Appellants Reply Brief at 22.
. See JA 2173-75; 2297-2303; 2569-71.
. Issacson v. Toyota Motor Sales, 438 F.Supp. 1, 7 (E.D.N.C.1976).
. See Sklar, “Second Collision” Liability: The Need for Uniformity, 4 Seton Hall L.Rev. 499, 527 (1973).
. See Trial Transcript p. 7759.
. Contrary to the majority’s suggestion, the Wilson case is cited only for the proposition that a jury should give cognizance to the full design of the automobile, including safety factors, in determining whether an automobile is crashworthy. There is no intent to adopt the Wilson court’s mitigation of damages analysis which limited evidence on the non-use of seat belts to the determination of damages. As the trial court made clear crashworthiness and mitigation of damages are two clearly distinct concepts.
. The very rationale for admitting evidence pertaining to Safety Standard 212, namely that although VW was not bound by the standard, it was a jury question whether adherence to the standard constituted reasonableness in design, see Trial Transcript 7754-55, should similarly dictate admission of the seat belt evidence. Although plaintiffs were not bound to wear seat belts, evidence of their availability and non-use would properly be factors for the jury to consider in assessing crashworthiness.
. See JA 2173-75; 2292; 2297-2303; 2569-71; 2575-76.
. It is still unclear whether the Fatal Accident Reporting System (FARS) data relied upon here, while perhaps relevant, are derived from accidents analogous to the one in issue. See Trial transcript 1595-1603. Nor is it certain that the Federal Motor Vehicle Safety Standard 212, 49 C.F.R. § 571.212 (1978) introduced in the proceedings below, was intended to apply at all to the Type II van. See Trial transcript 1603, 7754-55.