Defendant McCabe Powers Body Company (McCabe) and third-party defendant Consolidated Edison Company of New York, Inc. (Con-Ed) appeal from a judgment of the District Court for the Eastern District of New York (Eugene H. Nicker-son, Judge) entered upon special verdicts, Fed.R.Civ.P. 49(a), holding appellants liable, and apportioning responsibility among them, for injuries suffered by plaintiff Thomas Cappellini in the course of his employment as a lineman for Con-Ed while operating equipment manufactured by McCabe. We conclude that the evidence was insufficient to support one of the two special verdicts imposing liability upon ConEd and that the elimination of this ground for Con-Ed’s liability requires redetermination of the defendants’ comparative responsibility for Cappellini’s injuries.
I. The Evidence and the Verdicts
Cappellini was injured on May 1, 1973, while working with several other Con-Ed linemen on a project in Staten Island, New York. The project involved removal of old electric poles and the wires they supported. To gain access to the wires, the linemen used hydraulically powered “buckets” manufactured by McCabe.
Before beginning work, Cappellini prepared his bucket for use. First, he started the truck that carried and supplied power to the hydraulic machinery from which the bucket, in turn, drew its power. He then reached inside the truck’s cabin and pulled out a button located on the dashboard. By pulling out the dashboard button, plaintiff activated a “kill switch” situated inside the bucket. When activated, this switch allowed the lineman using the bucket to cut off the power generated by the truck below and thereby stop the movement of the bucket in an emergency.
After testing the kill-switch to make sure that it was activated, Cappellini entered the bucket and caused it to rise toward the upper part of the pole on which he was working. He controlled his speed and direction by manipulating levers in a control box on the outer side of the bucket’s upper *3rim. The control box was surrounded on three sides by a fiberglass “safety guard” that rose above, but not over, the levers.
Cappellini first removed the pole’s lower cross-arms, and then directed the bucket upwards. As the bucket traveled toward the higher cross-arms, the control box came into contact with a horizontal cable. Notwithstanding the protection afforded by the fiberglass guard, the cable jammed his hand against the operating levers, preventing use of those levers to stop the bucket’s upward course.
Cappellini sought to stop the bucket by using the kill switch, but the switch did not function. Nor was the bucket’s movement halted by the “relief valve,” an alternate safety mechanism designed to cut power off automatically when the pressure against the bucket exceeded a specified level. Instead, the bucket’s power continued unabated, and the upward pressure snapped the cable that was pressing down on the control box. When the cable snapped, Cappellini was thrown from the bucket. He sustained severe injuries as a result of his forty-foot fall.
Cappellini and his wife subsequently sued McCabe in this diversity action, alleging that both the fiberglass safety guard and the kill-switch system had .been defectively designed and that the relief valve had been defectively manufactured. Plaintiffs contended that the safety guard was defective because it did not form a hood over the bucket’s control levers and that the kill-switch system was defective because, by requiring its operating button to be pulled out for activation, the system was susceptible to being turned off inadvertently if someone brushed against the button. McCabe denied these claims, raised the defense of contributory negligence, and impleaded Con-Ed as a third-party defendant. In denying its liability based on defective design of the safety guard, McCabe contended not only that its design was adequate but also that any design defect of the guard was not a cause of the accident. The cause, McCabe said, was Con-Ed’s conduct in cutting down the sides of the safety guard, thereby permitting Cappellini’s hand to become jammed against the controls by an overhead cable. As a third-party plaintiff, McCabe contended that Con-Ed negligently maintained the bucket and bucket truck. Among the instances of negligent maintenance by Con-Ed inferable from the evidence were Con-Ed’s cutting down the sides of the safety guard, allegedly done to permit the bucket truck to be garaged for repairs, and Con-Ed’s failure to assure that the relief valve was kept at a proper setting so that it would shut off power when pressure was created by the bucket’s movement against an obstruction. McCabe also contended that the “specifications” accompanying Con-Ed’s purchase order had dictated the particular design features at issue and that any defect in the design of the safety guard or the kill-switch system must therefore be attributed to the negligence of ConEd.
The jury was asked to answer a set of special interrogatories. Interrogatory 1, focusing on the safety guard and the kill-switch system, asked whether “a defect in the design” of the McCabe bucket truck was a substantial factor causing plaintiff’s injuries. The jury answered “yes.” Interrogatory 2, focusing on the relief valve, asked whether “a defect in the manufacture” of the McCabe bucket truck was a substantial factor causing plaintiff’s injuries. The jury answered “no” to this interrogatory and to Interrogatory 3, which asked whether Cappellini had been contributorily negligent.1 In response to Interrogatories 4 and 5 concerning the damages suffered by plaintiffs, the jury awarded $1,370,000 to Cappellini and $350,000 to his wife. '<
The jury was also asked to answer three additional interrogatories concerning *4McCabe’s third-party claim against Con-Ed. Interrogatory 6 asked whether a failure by Con-Ed “to exercise reasonable care in maintaining” the bucket truck was a substantial factor causing plaintiff’s injuries. As far as we can tell, the jury’s consideration of McCabe’s allegation that Con-Ed was liable on the third-party claim for negligent maintenance was not narrowed to any particular aspect of Con-Ed’s conduct.2 Focusing on the safety guard and the kill-switch system, Interrogatory 7 asked whether a failure by Con-Ed “to exercise reasonable care in the specifications it provided” was a substantial factor causing plaintiff’s injuries. The jury answered “yes” to both interrogatories. Interrogatory 8, to be answered if Interrogatory 6 or 7 was answered “yes,” asked the jury to determine “what percent of plaintiffs’ damages is attributable to the negligence of Con-Ed.” The jury answered “85%.”
The jury was not asked to allocate ConEd’s 85% share of responsibility between the two separate grounds of Con-Ed’s negligence. The judgment awarded plaintiffs $1,720,000 from McCabe, plus $25,870.68 interest from verdict to judgment, and awarded McCabe 85% of $1,745,870.68 from Con-Ed. Both defendants appeal.
McCabe contends that plaintiff was contributorily negligent as a matter of law and that the evidence was insufficient to permit the jury to find that either of its alleged design defects caused plaintiff’s injuries. Con-Ed, conceding that the evidence permitted a finding of its negligent maintenance (Interrogatory 6), contends that there is no evidence to support the jury’s finding (Interrogatory 7) that Con-Ed’s specifications caused McCabe to design the allegedly defective features of the safety guard and the kill-switch system. Con-Ed further contends that rejection of the “specifications” ground of its liability on the third-party complaint necessitates a new trial on the issue of comparative responsibility for Capellini’s injuries. We examine these contentions in turn.
II. McCabe’s Appeal
1. Contributory Negligence
McCabe first contends that, since Cappellini admitted that he was aware of the location of the horizontal cable, he failed, as a matter of law, to exercise reasonable care that would have avoided the accident. This argument overlooks the evidence, introduced by plaintiffs, from which the jury could properly have inferred both that Cappellini’s task required delicate maneuvering of the bucket and that the bucket’s controls made such maneuvering difficult. The availability of these inferences entitled the jury to conclude that Cappellini, even though aware of the danger overhead, was not shown to have failed to use reasonable care in operating the bucket.
McCabe’s second argument is based on an Occupational Safety and Health Administration (OSHA) regulation requiring persons working from an “aerial lift” to wear a “body belt” attached to the lift. It is conceded that Cappellini was not wearing such a belt. McCabe asserts that this violation of the OSHA regulation constitutes contributory negligence as a matter of law. Under New York law, however, *5which applies to this diversity action, a regulatory violation does “not establish negligence per se, but [merely constitutes] some evidence of negligence, which the jury could take into consideration with all other evidence bearing on that subject.” Schumer v. Caplin, 241 N.Y. 346, 351, 150 N.E. 139,140 (1925); Conte v. Large Scale Development Corporation, 10 N.Y.2d 20, 29, 176 N.E.2d 53, 57,217 N.Y.S.2d 25, 29 (1961). Cappellini’s failure to wear a body belt did not bar recovery.
2. Design Defects
(a) Kill-Switch System. Implicitly conceding that the jury could properly have found that the kill-switch was too easily deactivated, McCabe nevertheless contends that it could not be found liable on this design defect claim because there was no evidence that this defect caused the accident. According to McCabe, the record contains no evidence that any lineman entered the cabin of Cappellini’s bucket truck while he was working in his bucket and consequently no evidence that a lineman’s inadvertent contact with the dashboard button caused deactivation of the system.
Plaintiffs admit that no one saw a lineman enter the cabin of Cappellini’s truck while he was working in the bucket. They contend, however, that the jury could properly have inferred such entry from evidence that linemen frequently borrowed supplies from the cabins of bucket trucks other than their own and that they followed this practice on the evening of the accident.3 According to plaintiffs, this evidence allowed the jury to infer that the safety switch was deactivated when a lineman, while borrowing supplies from the cabin of Cappellini’s truck, inadvertently brushed against the dashboard button.
The decisive consideration in assessing the sufficiency of this evidence of causation is the “relative probability” of “possible explanations” for the accident, and “the legal inferences that can most reasonably be drawn from the most probable ones.” Evans v. S.J. Groves & Sons, 315 F.2d 335, 343 (2d Cir.1963). The only possible causes of the safety system’s failure to function could have been Cappellini’s failure to activate it, mechanical malfunction, or inadvertent deactivation. The jury was entitled to credit Cappellini’s claims that he pulled out the activation button and ascertained, by testing the kill-switch, that the system was functioning properly. By crediting these claims, the jury would have excluded the possibility of Cappellini’s failure to activate and would have substantially diminished the probability of mechanical malfunction. That would have left plaintiffs’ hypothesis of inadvertent deactivation by someone in the truck as a substantially probable cause of the system’s failure to function. Testimony that linemen frequently entered each others’ trucks reinforced that hypothesis, and, since we must view the evidence most favorably to the plaintiffs, Park v. Village of Waverly, 457 F.2d 1139 (2d Cir.1972); Hannon v. Schmitt, 18 A.D.2d 854, 236 N.Y.S.2d 107 (3d Dep’t 1963),4 we hold that plaintiffs’ theory was properly available for the jury to adopt.
*6(b) Fiberglass Safety Guard. McCabe contends that the evidence established as a matter of law that Con-Ed employees had cut down the fiberglass safety guard and that, as a result, any design defect in the safety guard could not have been the cause of Cappellini’s injuries. In support of this claim, McCabe relies on the testimony of Butters, a Con-Ed employee who supervised Cappellini’s work. Butters testified that he “knew” that Con-Ed cut down the sides of the safety guards by three to four inches to permit the bucket trucks to enter the garage where they were repaired. McCabe argues that this testimony was uncontroverted and was therefore required to be credited.
The evidence strongly supported McCabe’s claim that Con-Ed employees had cut down the sides of the safety guard and surely permitted the jury to accept McCabe’s argument that its defective design of the safety guard was not the cause of the accident, in light of Con-Ed’s subsequent modification. If the jury so concluded, it still was entitled to hold McCabe liable for defective design of the kill-switch. But the jury was not obliged to accept Butters’ testimony. Butters conceded that he “did not see the shields [guards] being cut down”5 and relied on his observation that “[s]ome shields were short, and some shields were long.” His testimony left open the question whether the safety guard on plaintiff’s bucket was one of those whose height had been reduced.6 Because Butters’ testimony only permitted but did not require a conclusion that the safety guard on Cappellini’s bucket was modified, McCabe’s last claim must be rejected.
III. Con-Ed’s Appeal
New York permits a defendant found liable for a plaintiff’s injuries to recover partial or even full indemnity from the plaintiff’s employer if the employer’s negligence was also a proximate cause of the injuries. Dole v. Dow Chemical Co., 30 N.Y.2d 143, 282 N.E.2d 288, 331 N.Y.S.2d 382 (1972); Westchester Lighting Co. v. Westchester County Small Estates Corp., 278 N.Y. 175, 15 N.E.2d 567 (1938). This result, which Westchester Lighting held was not barred by the insulation that New York’s Workers’ Compensation law affords the employer against suit by the employee, is said to arise from an “independent duty or obligation” owed by the blameworthy employer to the blameworthy defendant. Westchester Lighting, supra, 278 N.Y. at 179, 15 N.E.2d at 568. Though the theory of an implied liability for indemnity appears to have developed in cases where it was equitable for the defendant to recoup all of its payment to the plaintiff from a third-party deemed primarily or actively liable for the injuries, Jackson v. Associated Dry Goods Corp., 13 N.Y.2d 112, 116-17, 192 N.E.2d 167, 169-70, 242 N.Y.S.2d 210, 213-14 (1963); McFall v. Compagnie Maritime Beige, 304 N.Y. 314, 330, 107 N.E.2d 463, 471-72 (1952), the New York Court of Appeals has ruled that a defendant may apportion its liability with a third-party defendant, including an employer, depending on *7“the proportion of the blame found against the third-party defendant.” Dole v. Dow Chemical Co., supra, 30 N.Y.2d at 153, 282 N.E.2d at 295, 331 N.Y.S.2d at 391.
With respect to McCabe’s third-party claim in this case, Con-Ed does not dispute its liability for a share of the damages that corresponds to its proportionate share of fault in negligently maintaining the bucket truck. However, it challenges the verdict assigning it 85% responsibility for plaintiffs’ damages on the ground that this apportionment comprises two claims of liability— negligent maintenance and negligent preparation of design specifications — and that the evidence was insufficient to permit the jury to find Con-Ed liable for design specifications. We agree with Con-Ed’s sufficiency argument,7 and conclude that the apportionment, grounded on a legally improper basis of liability, must be redetermined.
McCabe had sought to persuade the jury that the specifications Con-Ed had prepared required McCabe to design and manufacture the bucket with both of the design features the plaintiffs alleged were defective. Regarding the safety guard, Con-Ed’s specifications provided only that “a fiberglass guard shall be placed over the basket controls.” No reasonable jury could fail to find that this specification left McCabe free to determine both the shape of the guard and how high the sides of the guard should extend above the operating levers. Consequently, no reasonable jury could attribute liability to Con-Ed for the alleged defect in the shield’s design.
Regarding the kill-switch system, the specifications did no more than request an “insulated air-operated kill-switch mounted in the [bucket].” Balogh, who designed the bucket for McCabe, conceded that the requested device was an “option” that was available to all customers and that it was conceived and designed prior to McCabe’s receipt of Con-Ed s purchasing order. Balogh acknowledged that McCabe determined the location of the activation button and decided that affirmative action — pulling out the dashboard button — would be required to activate the kill-switch. Balogh’s uncontroverted admissions precluded the jury from finding that Con-Ed had any liability for design of the kill-switch system.
McCabe seeks to avoid the force of Balogh’s admissions by arguing that, even if Con-Ed did not act as a designer, it nonetheless “adopted” and assumed legal responsibility for McCabe’s design when it decided, after inspection, to take delivery of the bucket trucks. This argument finds no support in the law of products liability. The doctrine of products liability is based on the theory that accident costs will be both minimized and most appropriately distributed if primary responsibility for reasonable product safety is placed on the party controlling manufacture, Codling v. Paglia, 32 N.Y.2d 330, 340-42, 298 N.E.2d 622, 626-28, 345 N.Y.S.2d 461, 467-69 (1973), or design, Littlehale v. E.I. du Pont de Nemours & Co., 268 F.Supp. 791, 802 n. 16 (S.D.N.Y.1966), aff’d, 380 F.2d 274 (2d Cir. 1967); Munger v. Heider Manufacturing Corp., 90 A.D.2d 646, 456 N.Y.S.2d 271 (3d Dep’t 1982), or on a seller farther down the distributional chain, Nickel v. Hyster Co., 97 Misc.2d 770, 771,412 N.Y.S.2d 273,274 (Sup. Ct., Suffolk Co., 1979), who may seek indemnity or contribution from the manufacturer or designer. Dole v. Dow Chemical Co., supra, 30 N.Y.2d at 148-49, 282 N.E.2d at 292, 331 N.Y.S.2d at 387. Contrary to this underlying rationale, McCabe’s “adoption” theory would place primary responsibility for product safety on those who exercise no control over design or manufacture, but merely purchase for use.
Having concluded that it was error to permit the jury to consider McCabe’s
*8claim that Con-Ed was liable for negligent design specifications, we next consider whether, as Con-Ed urges, that error requires retrial of the apportionment of liability between McCabe and Con-Ed. Normally, a jury verdict may not stand if the jury was erroneously permitted to base its verdict on an invalid ground. See New York Times Co. v. Sullivan, 376 U.S. 254, 274,284, 84 S.Ct. 710,722, 728,11 L.Ed.2d 686 (1964); Stromberg v. California, 283 U.S. 359, 367-68, 51 S.Ct. 532, 535-36, 75 L.Ed. 1117 (1931). Since the jury was permitted to consider the invalid ground of negligent design in addition to the valid ground of negligent maintenance in determining that Con-Ed bore 85% of the responsibility for plaintiffs’ damages, Con-Ed assumes that jury redetermination, without consideration of the invalid ground, is required. This aspect of the appeal merits further consideration.
The rule requiring retrial when a verdict might rest on an invalid ground is regularly applied to cases in which the jury was asked to return a general verdict. One of the virtues of special interrogatories is that their use frequently permits a judgment to be modified, without the need for retrial, in the event that one aspect of the jury’s decision-making, reflected in an interrogatory response, is determined to be legally deficient. To maximize that benefit the parties and the trial judge must carefully design the interrogatories so that claims of uncertain validity are isolated for the jury’s discrete consideration. We do not mean to imply that special interrogatories must routinely be drawn so as to focus jury fact-finding on narrowly defined issues. Excessive refinement of interrogatories might needlessly complicate the jury’s task and discourage use of the normally helpful interrogatory technique. Our point is simply that, when a claim of doubtful validity is submitted to a jury, it will frequently be useful to frame an interrogatory that not only secures the jury’s resolution of the issue presented by the claim but also isolates the significance of its response from the other issues in the case. In this way a court that has submitted a doubtful claim to a jury to avoid the risk that its omission will lead to a reversal and a retrial can often avoid the competing risk that its inclusion will have the same consequences.
In this case the District Judge made good use of interrogatories by eliciting from the jury its separate assessment of numerous aspects of the claims made by the plaintiffs and by the third-party plaintiff. However, one respect in which specificity was not achieved is now significant. Though the jury was asked to determine separately whether Con-Ed contributed to Cappellini’s injuries by negligent preparation of design specifications and by negligent maintenance, it was not asked to determine a degree of responsibility attributable to each aspect of Con-Ed’s conduct. Perhaps the District Judge thought the jury’s task was sufficiently difficult in determining an allocation between McCabe’s conduct and ConEd’s aggregate conduct. Nevertheless, we face the issue whether Con-Ed, having raised no objection to Interrogatory 8, which did not seek separate percentages for each aspect of Con-Ed’s alleged negligence, is necessarily entitled to a new trial on apportionment once its objection to submission of the negligent specifications claim is sustained on appeal.
Our resolution of this issue will be aided by consideration of the circumstances that would have existed if Con-Ed had requested a more precise interrogatory that identified the share of responsibility attributable to each aspect of its conduct.8 Let *9us assume that the jury, if asked to divide Con-Ed’s 85% share of responsibility into two parts, had attributed, for example, 75% responsibility to its negligent maintenance and 10% to its negligent specifications, with McCabe bearing the remaining 15% share of responsibility (as the jury in fact found). This allocation would reflect the jury’s view that design defect, for which it held both Con-Ed and McCabe responsible, was a 25% cause of the accident (10% attributable to Con-Ed’s specifications plus 15% attributable to McCabe’s design). Once it is determined on appeal that Con-Ed bears no responsibility for design defect, a plausible resolution of the dispute, “just under the circumstances,” 28 U.S.C. § 2106 (1976), would be to direct the entry of a judgment that assigns 25% responsibility to McCabe and 75% to Con-Ed.9 Thus, a more precise interrogatory would have provided a permissible, though not necessarily a required, alternative to a new trial on the issue of apportionment.
The question then becomes what should be the consequence of Con-Ed’s failure to request a more precise interrogatory. It is arguable that the result should be affirmance of the 859^15% apportionment the jury made on the theory that Con-Ed waived the opportunity to remedy the error in submitting this ground of liability to the jury. Though we find this result tempting, we conclude that it is unwarranted. It was McCabe that sought to have a negligent specifications claim put to the jury, and Con-Ed duly objected. Though Con-Ed could then have sought a more precise interrogatory to isolate the effect of the error about to be committed, no rule or precedent required it to do so. We do not think it should be required to pay a judgment that includes some amount of dollars attributable to a claim that was erroneously submitted to a jury over its objection. The use of more precise interrogatories remains an available technique for parties and trial judges to use so that the risk of retrial may be reduced. But in this case the absence of an interrogatory separately identifying the share of responsibility attributable to each of the two grounds of Con-Ed’s liability cannot be used to deny Con-Ed a retrial on the issue of apportionment. Upon retrial of *10that issue, the jury should be informed that McCabe has been determined to be liable for design defect, that Con-Ed has been determined to be liable for negligent maintenance, and that the conduct of both has been determined to have been a substantial cause of the accident. The jury need determine only the apportionment of responsibility between McCabe and Con-Ed.
Accordingly, we affirm that portion of the judgment in favor of the plaintiffs against McCabe, reverse that portion in favor of McCabe against Con-Ed, and remand for further proceedings, consistent with this opinion, limited to retrial of the issue of apportionment.
. All parties concede that the doctrine of contributory negligence remained applicable when plaintiffs’ cause of action accrued. On September 1, 1975, New York adopted comparative negligence in actions to recover damages for personal injury. 7B N.Y.Civ.Prac.Law § 1411 (McKinney 1976).
. Interrogatory 6 refers only to “maintaining” the bucket, without any particularization, and the District Judge’s charge simply informed the jury of McCabe’s claim that Con-Ed “was negligent in its maintenance of the bucket truck.” It is arguable that the lawyers had in mind narrowing the negligent maintenance claim to Con-Ed’s alleged failure to keep the relief valve at a proper setting. In pre-charge colloquy counsel for Con-Ed argued that his client’s negligent maintenance was not an issue in the case; counsel for McCabe replied, “With respect to the relief valve it is.” This limitation may have been in the District Judge’s mind when he framed Interrogatory 6 to be answered only if the jury answered “yes” to Interrogatory 1 and “no” to Interrogatory 2, which he told the jury concerned only the relief valve. He may have thought that if McCabe was found liable for manufacturing a defective relief valve, this liability would be exclusive of any liability Con-Ed might otherwise have for negligent maintenance of the relief valve. On the other hand, when the District Judge rejected Con-Ed’s post-trial motion to set aside the finding of liability for negligent maintenance, he relied solely on evidence that Con-Ed either cut down the safety guard or failed to replace a safety guard that had been removed prior to the accident.
. It is fairly inferable from the record that the number of linemen working in Cappellini’s vicinity when the accident occurred was no less than eight and might have been as high as twelve.
. The New York cases concerning sufficiency of circumstantial evidence to establish negligence or proximate causation are not entirely consistent. Some decisions have stated that circumstantial evidence will support a verdict only where the inference sought to be drawn from that evidence “is the only one that is fair and reasonable.” Ridings v. Vaccareilo, 55 A.D.2d 650, 651, 390 N.Y.S.2d 152, 154 (2d Dep’t 1976); Boyce Motor Lines v. State, 280 A.D. 693, 117 N.Y.S.2d 289, 291 (3d Dep’t 1952), aff'd, 306 N.Y. 801, 118 N.E.2d 819 (1954). Other decisions have adopted a more lenient standard, holding circumstantial evidence to be sufficient as long as the inference sought to be drawn is “reasonable,” even though contrary inferences might also reasonably be drawn. Wragge v. Lizza Asphalt Construction Co., 17 N.Y.2d 313, 319-21, 217 N.E.2d 666, 670-71, 270 N.Y.S.2d 616, 620-22 (1966); Dillon v. Rockaway Beach Hospital & Dispensary, fas expressed in the first group of cases would drastically limit the use of circumstantial evidence, approaching the standard appropriate for a plaintiffs directed *6verdict. It seems doubtful that New York would so constrict the opportunity of plaintiffs to rely on circumstantial evidence of causation. We note that in most of the cases expressing the stringent rule the circumstantial evidence would not have been sufficient even under the lenient rule. We are of the view that the lenient rule, allowing the jury to choose among all inferences that might reasonably be drawn from the evidence, is the most appropriate expression of New York law. See Eldred v. Town of Barton, 505 F.2d 186, 188 (2d Cir.1974).
. The terms “shield” and “guard” were used interchangeably throughout the trial.
. Butters claimed that a few days prior to trial Cappellini called him and requested that he testify that Con-Ed did not cut down the shields and “that the shield was the full height.” While this testimony permitted the jury to find that Cappellini’s shield had been modified, it did not require such a finding because Cappellini denied Butters’ account of their telephone conversation, and the jury was entitled to credit Cappellini’s testimony over that of Butters. No other probative evidence was introduced on this issue. Butters expressly conceded that he could not recall observing the shield on Cappellini’s bucket on the day of the accident. Physical examination of the shield was impossible because neither McCabe nor Con-Ed was able to locate the shield after the accident occurred.
. As an alternative to its evidentiary argument, Con-Ed contends that section 11 of New York’s Workers’ Compensation Law (McKinney 1965 & Supp.1982) not only insulates it from a direct suit by plaintiffs but also precludes the imposition of third-party liability on a claim that ConEd participated in the design of the bucket trucks. Because we find the evidence insufficient to support the third-party “specifications” verdict, we need not assess this alternative argument.
. We are assuming that under New York law the apportionment to be made when a defendant seeks partial indemnity from a third-party defendant should reflect, primarily if not exclusively, the relative causal significance of each defendant’s conduct. “The primary inquiry in any such case should be the extent to which each of the negligent parties has contributed to the defective condition.” D’Ambrosio v. City of New York, 55 N.Y.2d 454, 464, 435 N.E.2d 366, 370, 450 N.Y.S.2d 149, 153 (1982). There is a suggestion in some of the New York cases that apportionment should also reflect the relative degrees of negligence of the two parties whose conduct caused the accident, as distinct from the relative causal significance of the two *9parties’ conduct. See Rogers v. Parise, 75 A.D.2d 513, 514, 426 N.Y.S.2d 750, 751 (1st Dep’t 1980) (using remittitur to revise jury’s determination of the extent to which a defendant was “more negligent” than a contributorily negligent plaintiff). See also Koster v. Fenton, 84 A.D.2d 783, 444 N.Y.S.2d 30 (2d Dep’t 1981). The standard New York jury instruction on apportionment appears to invite the jury to consider degrees of negligence, though we suspect that most juries take the instruction as a request to apportion the causal effect of the parties’ conduct. The instruction reads:
In making [the apportionment] you will consider under the instructions already given you the duty owed by each defendant to the plaintiff and the extent to which the conduct of that defendant deviated from the duty that he owed and weigh the relative degree of fault of each, expressed as a percentage of the total fault of both.
Weighing all the facts and circumstances, you must decide what constitutes a fair apportionment of the responsibility of each defendant for plaintiffs injury. In your verdict you will state the percentage of fault attributable to each defendant and the total of those percentages must add up to 100.
1 New York Pattern Jury Instructions — Civil § 2:275 (1974) (emphasis added).
In this case the jury was told, without objection, to determine “what percent of liability is attributable” to Con-Ed and “what constitutes a fair apportionment of responsibility of McCabe and Con-Ed” for the accident. This language is properly understood to focus on the defendants’ relative responsibilities for causing the accident, rather than their relative degrees of deviation from a standard of reasonable care. Indeed, it would be anomalous to focus a jury’s attention on relative degrees of negligence in a case such as this in which McCabe’s liability is grounded upon strict products liability and not negligence.
. Alternatively, it might be argued that if a jury apportions responsibility among three aspects of conduct and one aspect is determined to be an invalid basis for imposing any liability, a proper resolution would be to apportion the entire responsibility according to the relative shares the jury assigned to the two aspects on which liability was validly based. In our example, this approach would assign 8373% to ConEd and 162/3% to McCabe. While that approach might have considerable merit if the three aspects of conduct were totally distinct, it seems less appropriate here since Con-Ed’s specifications and McCabe’s design were both found by the jury to be causes of the defective design of the same product.