This is a negligence suit under the Longshore and Harbor Workers Compensation Act ("LHWCA"). An injured longshoreman sued the ship on which he was injured.1 A jury found for the defendant ship, and the plaintiff, assigning several errors in the trial, asks us to vacate the judgment of the District Court and remand for a new trial. For the reasons that follow, we will do so.

Under the LHWCA, injured longshoremen are barred from suing their employers, the stevedoring companies that contract with shipowners for loading and unloading work. 33 U.S.C. § 905(a). Instead, the stevedoring companies pay statutory compensation to injured longshoremen. 33 U.S.C. § 904(a). Longshoremen are, however, permitted to bring negligence actions against the ship on which they were injured. 33 U.S.C. § 905(b).

Plaintiff Cornelius Hill was injured while unloading cargo in the hold of defendants' ship, the Sea Panther, on August 24, 2000. He and another longshoreman, one Dwight Jones, were loosening the steel "lashing rods" which hold the cargo containers in place. While Jones was attempting to loosen a rod, it sprung off its housing and flew through the air, hitting Hill in the head, smashing his hard hat, knocking him unconscious and almost killing him. App. 63-64.

The lashing rods are thin steel rods, threaded at the ends. They are attached to the deck, or to the top of a container (containers are stacked several layers deep), and then to the corners of each container, where they are screwed tightly into turnbuckles. Turnbuckles are threaded cylinders into which the rods are inserted and then tightened with wing nuts. The rods, when tightened, are under enormous tension, and if a turnbuckle or rod is rusty or improperly installed, it can weaken, break or come loose.

Jones testified that the rod, nut, and turnbuckle were rusty and improperly installed, and that the turnbuckle was several inches out of place and was "frozen" on the rod. App. 63, 68-70. Jones said he saw rust on the turnbuckle, and no grease. Properly maintained rods, turnbuckles, and nuts,2 are regularly greased, to prevent rusting. Jones was unable to turn the wing nut, so, in accordance with customary longshoremen's practice, he struck it with his wrench to loosen it. App. 63. When he struck the wing nut, the rod snapped off its housing and flew through the air some thirty-two feet before hitting Hill. App. 64.

While at sea, a ship's crew must continually inspect the lashing assembly, because if cargo containers shift or fall, they can unbalance and even sink the ship. App. 216-17. The ship's captain testified that the crew did so here. App. 295. Upon docking, the crew "turns over" the ship to the stevedoring company for unloading. The ship has a legal duty to turn over the ship to the longshoremen in safe condition for unloading. Scindia Steam Navigation Co. v. De Los Santos, 451 U.S. 156, 167, 101 S. Ct. 1614, 68 L. Ed. 2d 1 (1981); Kirsch v. Prekookeanska Plovidba, 971 F.2d 1026 (1992). Hill alleged that the lashing assembly that injured him was rusty and improperly installed, and that the ship breached its turnover duty by leaving the rusty, improperly installed lashing in place without repairing it or warning the stevedore.

In response, the ship asserted two theories. First, it contended that if the turnbuckle was rusted, any hazard that it created should have been open and obvious to Jones, and that Jones was negligent in hitting the rusted turnbuckle with his wrench rather than seeking help from his supervisor. Second, the ship's expert hypothesized that the accident had not in fact taken place as Jones testified. Rather, the expert suggested, Jones might have partially loosened the turnbuckle without fully loosening the wingnut, thereby causing the turnbuckle to jam. Then when Jones struck the partially loosened turnbuckle he failed to hold on to the lashing rod, causing the rod to spring free of its casing. Thus, the ship claimed, Jones's actions, not the ship's, were the cause of the accident.

The case went to trial and a jury found in favor of the ship. Hill moved for a new trial, and now appeals from the denial of that motion. He raises three objections to the jury instructions and one to the admission of expert testimony. He argues that the District Court's instruction on superseding cause was in error, that the District Court misstated the law as to the ship's turnover duty, and that he was entitled to a res ipsa loquitur instruction. The testimony to which he objects is that of the defense's expert witness, who testified that it would have been physically impossible for the accident to occur in the way that Jones claimed it did. Hill argues that in so testifying, the expert went outside the bounds of his written report, and that Hill suffered from unfair surprise.

We begin with the ship's "turnover duty." Hill contends that the jury instructions were an inaccurate statement of the duty as explained most recently by this Court in Kirsch, 971 F.2d 1026. Hill requested an instruction that the ship would have a duty to fix or warn about the turnbuckle if the ship should have known that the longshoremen would not be able to ameliorate it by "practical" measures. The District Court declined to give that instruction. Our review of the legal correctness of jury instructions is plenary. Parks v. AlliedSignal, Inc., 113 F.3d 1327, 1330 (3d Cir. 1997).

The defendants do have a duty to warn of latent defects in the cargo stow and cargo area. This duty is a narrow one and attaches only to latent hazards defined as hazards that are not known to the stevedore and that would be neither obvious to nor anticipated by a skilled stevedore in the competent performance of his work. The duty encompasses only those hazards that are known to the vessel or should be known to it in the exercise of reasonable care.

As I mentioned above, the defendants are not liable if the danger that caused Plaintiff Hill's injuries would have been obvious to a reasonably competent stevedore. However, there is an exception to this rule. The defendants may be liable for an obvious hazard because custom, positive law, or contract instructs the ship owner to rectify the particular hazard, regardless of its obviousness.

For example, where a ship owner should know that longshore workers frequently confront rather than avoid a type of obvious hazard, the ship owner may be negligent in not limiting the hazard.

App. 445.

Translating appellate opinions into jury instructions is a notoriously difficult undertaking, and we take note of the fact that the District Court clearly read and attempted to apply Kirsch. Indeed, some of the quoted language is taken almost verbatim from Kirsch, see, e.g., 971 F.2d at 1026 ("On the other hand, customary practice may suggest that the shipowner should know that longshore workers frequently confront rather than avoid a type of obvious hazard. If so, the shipowner may be negligent in not eliminating the hazard . . ."). However, our review of the legal correctness of jury instructions is plenary, and, mindful of the fundamental importance to LHWCA cases of precise articulation of the turnover duty,3 we are compelled to conclude that the instruction given by the District Court did not accurately state the law as set out in Kirsch.

Kirsch's statement of the law is as follows:

[A] shipowner can, ordinarily, reasonably rely on the stevedore (and its longshore employees) to notice obvious hazards and to take steps consistent with its expertise to avoid those hazards where practical to do so. ... [A] shipowner may be liable for failing to eliminate an eliminable hazard only if it should have expected that its expert stevedore would not avoid the hazard and conduct cargo operations safely.

971 F.2d at 1031, 1033 (emphasis added). The highlighted language is crucial to the instructions here. There are two components to the rule on open and obvious hazards under Kirsch. When a ship is turned over to the stevedore with an open and obvious hazard which injures a longshoreman, the ship will be liable, first, if avoiding the hazard would be impractical for the longshoreman, or second, if the ship should have known that the longshoremen would confront the hazard. The District Court's instruction conveys only the second half of this rule. Thus it does capture the situation we described in Kirsch when we said that

Kirsch would be able to defeat summary judgment if he could offer evidence that, in light of custom . . . at that port or in this industry, [the owner] would have acted unreasonably to assume that [the] workers would avoid the danger, . . . that stevedores and longshore workers frequently proceed with cargo operations in holds despite large oil slicks there, which might imply that [the owner] should have expected that they would do so here.

Id. at 1034.

Just as in Kirsch the shipowner would have had a duty to warn of or mitigate the oil slick if the shipowner reasonably should have known that longshoremen regularly walk through oil slicks, so too, in this case, if the shipowner reasonably should have known that longshoremen regularly confront the hazard of rusty or misaligned turnbuckles, the shipowner would have a duty to mitigate or warn of rusty or misaligned turnbuckles. In this regard the jury was adequately instructed.

While we do not find error in the District Court's instruction on confrontation of hazards, we will note by way of guidance that we find the introductory phrase "for example" confusing as the District Court used it in the instruction, because it implies that the duty to rectify hazards which the shipowner reasonably should know that longshoremen regularly confront is a specific example of the general category of duties in which "custom, positive law, or contract instructs the ship owner to rectify the particular hazard." But the duty to mitigate regularly confronted hazards (reasonably known to be such) is not simply an instance of the duty to mitigate hazards which the shipowner is required to rectify by custom, positive law, or contract. Instead, it is a separate and independent duty, and it is grounded solely in the knowledge that a reasonable shipowner would have about longshoremen's customary practices. This is to say, regardless of what custom, positive law, or contract independently have to say about the ship's duties, our common maritime law finds duties where longshoremen regularly confront hazards and the ship should reasonably be aware of that practice. It is not entirely clear to us that a reasonable jury would so understand the instruction as given.

We do find error in the District Court's refusal to include the requested charge on "practical measures." To be sure, the hazards and habits encompassed by the phrases "frequently confront" and "cannot avoid by practical measures" may overlap to some degree. But they are not identical. A particular rarely-occurring hazard may be impractical to avoid, so that it would not be the case that longshoremen frequently confront it, or that a reasonable shipowner would know that they do. Nonetheless, if the hazard cannot practically be avoided, the shipowner may have a duty under Kirsch to mitigate it. And conversely, as we said in Kirsch, "there may be cases where the shipowner cannot reasonably expect that [longshoremen] will avoid an obvious hazard even when practical to do so." 971 F.2d at 1030-31.

The "practical measures" duty has nothing to do with the shipowner's knowledge, or with the frequency of occurrence of the hazard, but simply with "whether, under all the circumstances, safer alternatives were impractical." Id. at 1030. This duty may attach even when there are in fact alternative courses of conduct available. In Kirsch, 971 F.2d at 1030, we quoted with approval the Fifth Circuit's formulation of the ship's duty in Morris:

[T]he longshoreman need not show that he had no possible alternative but to use defective equipment or to work in a dangerous area. The burden is not so heavy. He need show only that the circumstances made safer alternatives unduly impractical or time-consuming.

Morris v. Compagnie Maritime Des Chargeurs Reunis, S.A., 832 F.2d 67, 71 (5th Cir. 1987). Morris, in turn, quoted with approval the Second Circuit's statement that a duty will attach when the longshoreman's "only alternatives would be to leave his job or face trouble for delaying the work." Napoli v. Transpacific Carriers Corp., 536 F.2d 505, 509 (2d Cir. 1976).

If the alternatives to striking a frozen wing nut with a wrench were impractical — if the longshoreman's only alternatives are to leave the job or face trouble for delaying the work — then the shipowner had a duty to warn of or mitigate the hazard created by such turnbuckles, even if it was open and obvious. If the evidence could reasonably have supported such a finding, then the instructions should have specified the existence of such a duty.

There was considerable testimony from both parties on longshoremen's options and practices when faced with frozen turnbuckles. That testimony concerned both the frequency and the practicality of various potential responses. For example, the defendant's expert witness, Walter Curran, suggested that Jones should have stopped work when he encountered the frozen turnbuckle and reported it to his superiors. App. 365. Jones, on the other hand, testified that a longshoreman who stopped work to report a frozen turnbuckle to his superiors would be fired as incompetent. App. 66. No other alternatives were proposed, so the jury could reasonably have concluded that reporting a frozen turnbuckle is an impractical way to avoid the risks of hitting the wing nut. Thus evidence was developed at trial which could have established a legal duty, but the jury was not told of the existence of that duty. This was error.

We must accordingly ask whether that error was harmless. "An error will be deemed harmless only if it is `highly probable' that the error did not affect the outcome of the case." Forrest v. Beloit Corp., 424 F.3d 344, 349 (3d Cir. 2005) (citing McQueeney v. Wilmington Trust Co., 779 F.2d 916, 924 (3d Cir. 1985)). In this case, the jury returned a finding of "no negligence." That finding might have been based on a determination that the ship had no duty to mitigate the hazard created by the frozen turnbuckle. The jury was not told that the impracticality for longshoremen of alternatives to confronting that hazard can create a duty owed by the ship to mitigate it. Given the evidence presented, we cannot say that it was "highly probable" that the result would have been the same had the jury been correctly instructed on the "practical measures" rule.

In McQueeney, we explained the rationale for keeping a relatively tight rein on harmless error determinations:

[B]road institutional concerns militate against increasing the number of errors deemed harmless. Although it is late in the day to pretend that all trials are perfect, perfection should still be our goal. Judge (now Chief Judge) Robinson put the point well: "The justification for harmless-error rules is singleminded; they avoid wasting the time and effort of judges, counsel and other trial participants. Other considerations enter into the picture, however, when we set out to ascertain what is harmless and what is not. Wisdom of the ages counsels against appellate erosion of the stature and function of the trial jury. Societal beliefs about who should bear the risk of error in particular types of proceedings deserve weight in decisions on harmlessness. Respect for the dignity of the individual, as well as for the law and the courts that administer it, may call for rectification of errors not visibly affecting the accuracy of the judicial process. And the prophylactic effect of a reversal occasionally might outweigh the expenditure of effort on a new trial." By maintaining a moderately stringent, though not unreasonably high, standard in civil as well as criminal cases, we preserve a strong incentive for the district courts to minimize their errors, and we thereby bolster the integrity of the federal judicial process.

McQueeney, 779 F.2d at 927 (quoting United States v. Burton, 584 F.2d 485, 512-13 (D.C. Cir. 1978) (Robinson, J., dissenting)).

We are mindful of the respect due to a jury verdict, and of the crowded dockets in our district courts. However, we are also mindful of a litigant's right to have full and accurate legal instructions given to the jury. The instructions here were incomplete, and the omission reasonably could have affected the outcome of the trial. In order to assist future District Courts in crafting turnover duty instructions, therefore, we think it advisable to restate the relevant turnover rules as developed in our caselaw.

1. The ship has a duty to turn the ship over to the longshoremen in safe condition for unloading.

2. That duty includes mitigating open and obvious hazards if the ship reasonably should know that longshoremen either (a) are likely to work through them rather than mitigating them, or (b) are unable to mitigate them through practical measures.

3. Facts about the actual practices of longshoremen are relevant to the determination of what the ship reasonably should expect the longshoremen to do, and what measures are practical.

The instruction in this case omitted a substantive element of the ship's duty, and based on the evidence presented at trial we cannot conclude that it is highly probable that the omission did not affect the outcome. The judgment must therefore be vacated.

We turn now to the vexing topic of superseding cause. The District Court's instruction on superseding cause was as follows:

Now, you may also find that an act of a third party caused the accident and superseded all other causes. Generally, this means that the act of a third party was so unexpected and out of the ordinary, that it supersedes any negligen[t] act or acts that may have come before it. If you find that there is such a superseding cause, any and all negligent acts that occurred prior to a superseding cause are not considered a legal cause of the harm to the plaintiff.

In this case, the defendants contend that the act of a third party, namely Dwight Jones, was a superseding cause of the accident. If the act of Dwight Jones was a superseding cause of the accident, the defendants are not liable for any damages that the plaintiff . . . sustained as a result of the accident.

You may find that the act of Dwight Jones was a superseding cause of the accident, only if you find that the defendants have proven the following by a preponderance of the evidence:

First, the defendants had no reason to know Dwight Jones would act as he did.

Second, a reasonable person would conclude that Dwight Jones' act was highly extraordinary.

And third, the act of Dwight Jones was either extraordinarily negligent or not a natural consequence of any act or failure to act by the defendants.

App. 442.

Hill argues that there was not sufficient evidence presented to support a finding of superseding cause, and that, at least as applied in this case, superseding cause is conceptually in tension with the remedial scheme set forth in the LHWCA. We conduct our review for sufficiency of evidence de novo, asking "whether, viewing the evidence in the light most favorable to the nonmovant [in this case the ship] and giving it the advantage of every fair and reasonable inference, there is insufficient evidence from which a jury reasonably could find liability." W.V. Realty Inc. v. Northern Ins. Co., 334 F.3d 306, 311 (3d Cir. 2003).

The superseding cause instruction implicates both the principles of maritime tort law as they have developed through caselaw, and the federal legislative scheme for compensating longshoremen's injuries. We conclude that the liability-allocation scheme created by Congress in the LHWCA requires us to apply heightened vigilance to superseding cause instructions in longshoremen's injury cases. We therefore provide some background on the cases interpreting both superseding cause instructions in maritime cases and the relevant provisions of the LHWCA.

The rule in maritime cases, codified in the LHWCA, is that where a ship's negligence causes injury to a longshoreman, the ship is liable for the full amount of the longshoreman's damages, reduced only by the percentage of damages caused by the longshoreman's own negligence. Edmonds v. Compagnie Generale Transatlantique, 443 U.S. 256, 266, 99 S. Ct. 2753, 61 L. Ed. 2d 521 (1979). The concurrent negligence of non-defendants, such as statutorily immune stevedoring companies, is irrelevant. Id.

In 1975, the Supreme Court decided United States v. Reliable Transfer Co., 421 U.S. 397, 95 S. Ct. 1708, 44 L. Ed. 2d 251 (1975), and adopted comparative fault for the first time in maritime law.4 The Court held that where plaintiff and defendant were both negligent, damages should be apportioned according to relative fault. Id. at 411.

In Edmonds, the Court had to decide whether the Reliable Transfer principle affected LHWCA cases in which a non-defendant third party, the stevedoring company, is partly at fault. Should the defendant's liability be reduced in that situation? The Court emphatically said "no." Edmonds reversed a Ninth Circuit ruling in a longshoreman's injury case in which the ship was adjudged 20% negligent, the longshoreman 10%, and the stevedoring company 70%. The Ninth Circuit had held that the comparative fault principles adopted in Reliable Transfer entailed that the longshoreman could only recover 20% of his damages from the ship. Not so, said the Court; such a result would contravene the longstanding rule that where the ship is negligent, the ship pays for the full amount of the longshoreman's injuries, other than those resulting from the longshoreman's own negligence. "[W]e are quite unable to distill from the face of the [amendment] any indication that Congress intended to modify the pre-existing rule that a longshoreman who is injured by the concurrent negligence of the stevedore and the ship may recover for the entire amount of his injuries from the ship." Edmonds, 443 U.S. at 266. The longshoreman himself was responsible for only 10% of his damages; thus, the Court held, he could recover 90% from any concurrent tortfeasor.

The Ninth Circuit's rule, the Court reasoned, would be unfairly burdensome to the longshoreman, because he is barred from suing the stevedore, and, as Congress recognized, the statutory compensation payable under the LHWCA will in many cases be substantially less than his actual damages. Id. at 269. In cases where the stevedore's negligence caused more damages to the longshoreman than is payable under the LHWCA, the stevedore will get a windfall, and a concurrently negligent shipowner will have to bear an added cost. The Ninth Circuit, and the three dissenters on the Court, thought this arrangement unfair; the Court replied that judicial sympathy for the ship comes at the cost of a pound of the longshoreman's flesh, and that is not what Congress provided when in its wisdom it adopted the LHWCA. Congress squarely faced the choice between full shipowner liability and diminished longshoreman recovery, and, after extensive debate, chose the former.

In 1972 Congress aligned the rights and liabilities of stevedores, shipowners, and longshoremen in light of the rules of maritime law that it chose not to change. . . . By now changing what we have already established that Congress understood to be the law, and did not itself wish to modify, we might knock out of kilter this delicate balance. As our cases advise, we should stay our hand in these circumstances.

Id. at 273.

The Court emphasized that Congress has chosen to specify how the costs of longshoremen's injuries are to be apportioned, and there is no question that Congress has the power to do so. Congress has made stevedores immune from suit in longshoremen's injury cases, and that leaves the ship as the sole defendant. Settled principles of tort law dictate that "[a] concurrent tortfeasor . . . is not relieved from liability for the entire damages even when the nondefendant tortfeasor is immune from liability." Id. at 260 n.8 (citing Restatement (Second) of Torts § 880).

Reliable Transfer was brought to the Court's attention again in 1996 in Exxon v. Sofec, 517 U.S. 830, 116 S. Ct. 1813, 135 L. Ed. 2d 113 (1996). Exxon, which had been adjudged the superseding cause of its own damages, asked the Court to rule that the Reliable Transfer comparative fault principle was logically incompatible with superseding cause. The Court declined to do so, but gave no guidance beyond the definitional observation that superseding cause has to do with causation and comparative fault has to do with damages. A chorus of commentators protested that Sofec failed to appreciate the conceptual tensions in the relationship between superseding cause and comparative fault.5

The theoretical reticence of Sofec is brought home in the lower courts in disputes over jury instructions. In the case at bar, it is urged that a superseding cause instruction had the effect, contra Edmonds, of placing the burden of a non-negligent longshoreman's injury on the longshoreman himself. The legal problem presented by this case is thus as follows. The Supreme Court in Edmonds explained that when a longshoreman is injured by the concurrent negligence of a shipowner and a stevedore, the shipowner is liable for the full damages.6 The Supreme Court in Sofec held that superseding cause remains a viable analytical category in maritime cases. Superseding cause instructions, then, may be given in maritime cases. Does it follow that they must be given? And do LHWCA cases require heightened vigilance because of the congressional cost-allocation scheme we are bound to implement? On these questions Sofec is silent, and today we answer them "no" and "yes." If we permit superseding cause instructions to be given in LHWCA cases such that ordinary stevedore negligence could operate to cut off the liability of concurrently negligent shipowners, we will eviscerate the liability-attribution framework Congress created in 1972. To avoid this result we must closely scrutinize the evidentiary basis for the instruction.7

The Sofec Court did not address, or even cite, Edmonds. Its analysis of the compatibility of superseding cause with maritime precedent was restricted to Reliable Transfer and the advent of comparative fault in maritime law. Perhaps because Sofec was not a LHWCA case, the Court did not see a need to consider the effect of its ruling on Edmonds and the LHWCA. Whatever the Court's reasons, we are unwilling to interpret Sofec's silence in such a way as to contravene Edmonds.

First, Sofec held only that "there is [no] repugnancy between the superseding cause doctrine, which is one facet of the proximate causation requirement, and a comparative fault method of allocating damages." 517 U.S. at 838. Neither Edmonds nor this case involves comparative fault. See Edmonds, 443 U.S. at 273 n.30 ("Further, the stevedore is not a party here and cannot be made a party, so [Reliable Transfer] is inapplicable."). Second, the introduction of comparative fault to maritime law in Reliable Transfer was a judge-made innovation; by contrast, the LHWCA is a statute. Sofec does not therefore foreclose the possibility — though we do not decide the question here — that superseding cause per se may be doctrinally incompatible with the LHWCA as adumbrated in Edmonds. Third, the issue decided in Sofec was "whether a plaintiff in admiralty that is the superseding and thus the sole proximate cause of its own injury can recover part of its damages from [other cause-in-fact] tortfeasors." Id. at 840 (emphasis added). Hill was not a cause of his own injury in any way. Whether this difference in equities amounts to a distinction in law is also, fortunately, not something we need to decide in this case.

Finally, Sofec involved plainly extraordinary facts quite unlike those at issue here. The defendant in Sofec was the manufacturer of a mooring system used for transferring oil from a tanker into a pipeline. The line securing the tanker to the pipeline broke in a storm as the ship was off-loading its cargo of oil. The defendant's alleged negligence went only to the design of the mooring line. However, the line's breaking did not damage the ship; instead it required the ship to maneuver and get to a safe position. The ship did maneuver and get to a safe position. After having reached safety, however, and several hours after the line broke, the captain decided to turn back toward shore. The captain had neglected, however, to have the ship's position fixed during the maneuvering, which he could easily have done and should have done. The ship ran aground and was damaged. The district court found that turning toward shore without knowing the ship's position was a highly extraordinary act, and that it was not a consequence of the mooring line breaking. The normal routine during any maneuvering is for the captain to have the ship's position continually plotted, and despite the attention given over to repairing the mooring line and fuel hoses, there were crew members available to do the plotting. Thus Sofec presents a clear case of superseding causation. See Sofec, 517 U.S. at 833-35. The case at bar, as we will explain, does not.

Absent any discussion of Edmonds in Sofec, we must assume that the two cases are reconcilable. And if Sofec did not overrule Edmonds, then Edmonds and not Sofec is the controlling case on LHWCA liability. Edmonds holds unambiguously that where both the shipowner and the stevedore are negligent, the shipowner is liable for the full award. To be sure, if a superseding cause intervenes between the ship's negligence and the longshoreman's injury, then the ship is not liable at all, because it is not a proximate cause of the injury. The devil, here as always, is in the details: when can a fellow longshoreman's action constitute a superseding cause?

Commentators have long warned about the dangers — in maritime as well as in tort law generally — of what Hill suggests happened here: that a superseding cause instruction might invite a finding of no liability on facts which would otherwise be straightforwardly amenable to either sole liability for concurrently caused injuries, or to comparative negligence analysis.8 This danger is particularly acute in LHWCA cases, because Congress has specifically provided for shipowner liability in cases of concurrent shipowner and stevedore negligence. Courts must be vigilant, therefore, when crafting jury instructions, to ensure that they do not undermine the governing statutory scheme.

The Restatement (Second) of Torts § 442 suggests six factors by which superseding causal acts can be picked out against the causal background. Relevant here are the first three:9 (a) whether the act brings about a harm different from the harm that the defendant's negligence would have caused; (b) whether the act is "extraordinary" rather than "normal" under the circumstances; and (c) whether the act is independent of any situation created by the defendant's negligence.

We can certainly imagine cases in which the actions of a fellow longshoreman could constitute a superseding cause so as to insulate a concurrently negligent ship from liability. For instance, if Jones had pulled out a gun and shot at Hill in an attempt to kill him, but missed, hitting an improperly secured lashing rod instead, causing it to break free and strike Hill, a superseding cause instruction would be appropriate. The obvious and extreme case of an intentional tort, though, sheds little light on the problem at hand. Moving closer to the instant facts, if Jones had attempted a radical and untried unfreezing technique, applying, say, a blowtorch or a chainsaw to the wing nut, we would probably see no error in a superseding cause instruction.

What distinction between such a case and ours? The test is whether the unfreezing method employed by the longshoreman was an "extraordinary" one. Upon this concept rests the validity of the instruction. An extraordinary act is one which is not done in the normal course of events. An extraordinary method of unfreezing a turnbuckle is one to which longshoremen do not ordinarily resort when faced with frozen turnbuckles. "Ordinarily" means regularly; as a matter of habit, custom, usual practice; everyday. The Restatement factors highlight the importance of the "everyday course of events" as a baseline for evaluating the defendant's negligence. If subsequent third-party acts are carried out in the way they normally are, then the harm caused by the combination of the third-party acts and the defendant's negligence is not different in kind from the harm that the defendant's negligence "on its own" caused, because "on its own" includes the "everyday course of events" in which the defendant's negligence is situated. Likewise with the second and third factors: to determine whether an act is "extraordinary" rather than "normal," we ask what the usual practices are in the contextual environment at issue, and to determine whether the effects of the third-party cause are "independent" of the defendant's negligence, we ask whether the defendant's act was likely, in the ordinary course of events in that particular context, to lead to the third-party act.

In this case, the defendant's alleged negligence was in the maintenance and inspection of the lashing rod. Negligent maintenance and inspection of the lashing rods can cause injury to the longshoremen who must work with them at unloading, leading to precisely the sort of accident that happened here. Jones's act was precipitated by the bolt's being frozen in place, which was a likely, ordinary, and foreseeable consequence of its being misaligned, uninspected, and poorly maintained.

The Restatement further specifies several circumstances in which a subsequent third-party act is not a superseding cause, even if negligent.

The fact that an intervening act of a third person is negligent in itself or is done in a negligent manner does not make it a superseding cause of harm to another which the actor's negligent conduct is a substantial factor in bringing about, if

(a) the actor at the time of his negligent conduct should have realized that a third person might so act, or

(b) a reasonable man knowing the situation existing when the act of the third person was done would not regard it as highly extraordinary that the third person had so acted, or

(c) the intervening act is a normal consequence of a situation created by the actor's conduct and the manner in which it is done is not extraordinarily negligent.

Restatement (Second) of Torts, § 447. In this case Hill argued that because longshoremen customarily hit stuck wing nuts with their wrenches, the shipowner should have realized that that might occur; that a reasonable person would not think it extraordinary for a longshoreman to hit a stuck wing nut, and that Jones's hitting the wing nut was a direct and normal consequences of the ship's alleged negligence in maintaining and inspecting the turnbuckle.

To determine whether Jones's act could, as a matter of law, have been found to be a superseding cause on the record before us, we must determine what that record discloses about the ordinary practices of longshoremen. The superseding cause instruction will be upheld only if the record contains evidence from which a reasonable jury might conclude that it would have been extraordinary for a longshoreman in Jones's position to strike the wing nut with his wrench.

Every witness who spoke on the subject testified that hitting frozen wing nuts to unstick them is common longshoremen's practice. App. 66 (testimony of Jones); App. 166 (testimony of Hill);10 App. 315 (testimony of defense witness Vagn Ejsing).11 The shipowners put on no rebuttal witness. Even the defense's expert, Walter Curran, when asked to describe typical unloading practices, testified that when lashing bolts are stuck, "hitting the locking nut with the spanner wrench is actually quite common."12

Jones's description of the customary practice was as follows.

—Okay, on the Philadelphia waterfront for as long as you've been working there, what has been the custom and practice for dealing with a frozen turnbuckle?

—You try it with your wrench first and you try to break it. If not, then you bang it. Again, you're gonna — trying to bang it in the direction to make it turn — in the direction you want it to go to turn, that's the direction you hit it in.

—And what has been the custom and practice on the Philadelphia waterfront with respect to reporting a frozen turnbuckle to a hatch boss.

—Well, again, I'm a hatch boss and if someone did that to me, they came back — that's let me know that they don't know what they're doing. So I'm probably not gonna hire this guy again. If there is any way I can get around hiring this guy, I will. That's a simple problem, no one I've ever seen has come back to any hatch boss — including me — and said, I've got a frozen turnbuckle, I can't — you'd probably get fired. I would have gotten fired that night.

App. 66.

On cross-examination, Jones was asked why he didn't warn Hill before striking the wing nut:

—Was there anything preventing you, when you saw the turnbuckle frozen, from walking four containers and saying, Neal, I've got a problem here? Neal I'm gonna whack the turnbuckle, look out?

—No, again—

—Was there anything that stopped you from doing that?

—There was nothing that stopped me from doing that, other than the same practice that I've done every other time.

App. 102.

The testimony, in sum, from both the plaintiff's and the defendants' witnesses, was that longshoremen commonly and ordinarily bang stuck wing nuts with their wrenches in order to loosen them. There was no testimony offered at any point by any witness indicating that banging stuck wing nuts was anything other than the common and ordinary practice of longshoremen.

If superseding cause can be legally found on these facts, then the distinction between concurrent negligence and superseding cause will have evaporated, and Congress's carefully planned cost-allocation scheme will have been upended, in contravention of Edmonds.

Further, the superseding cause instruction here was capable of confusing and misleading the jury.13 The jury form asked, first, whether each defendant was negligent, and the jury answered "no." The second question, to be answered only if either defendant was found negligent, read: "If so, did the negligence, in whole or in part, cause any injury or damage to plaintiff Cornelius Hill?" The District Court held, and the ship urges on appeal, that this verdict cures any possible error in the superseding cause instruction. We do not agree.

The instruction stated: "[Y]ou may also find that an act of a third party caused the accident and superseded all other causes. Generally this means that the act of a third party was so unexpected and out of the ordinary, that it supersedes any negligent act or acts that may have come before it." App. 442. The jury returned a finding of "no negligence." It is possible, of course, that that finding was independent of any evaluation of Jones' act as a superseding cause. But that possibility is too speculative to support a harmless error determination, given the degree to which the defense's case focused on Jones. We think it amply possible that the jury focused on Jones, too, and interpreted the instruction as requiring a finding of "no negligence" for the ship if the jury found that both Jones and the ship were negligent.14

For purposes of harmless error analysis, therefore, we ask whether it is highly probable that the error did not affect the result, which means, in this case, whether there is a reasonable possibility that the jury interpreted the instruction to mean that Jones's negligence wiped out the ship's negligence so that Question One should be answered "no." We think there is. After all, one of the ordinary meanings of "supersede" is "to make void, to annul." Webster's Third, at 2295. We think it too much to insist on the presumption that the jury understood that if both Jones and the ship were negligent, and it found Jones to be a superseding cause, the ship's negligence would still be negligence ("yes" to Question One) but would no longer be a legally operative cause of Hill's injuries ("no" to Question Two). We think it just as likely that the jury understood the words "supersede any negligent act that came before it" to mean that they could find that Jones's act "voided" and "annulled" the ship's negligence, thus yielding the form they in fact returned ("no" to Question One). We do not think it "highly probable" that the form does not express a superseding cause finding. Therefore the form does not render the error in the instruction harmless.

The likelihood that the jury may have adopted the interpretation suggested above is heightened by the absence of any evidence of the "extraordinariness" of Jones' act. That absence is complete, and therefore, as to superseding cause, Hill is entitled to judgment as a matter of law on this record, because viewing the evidence in the light most favorable to the ship, and giving the ship the advantage of every fair and reasonable inference, there is insufficient evidence from which a jury could reasonably find" superseding cause. See Wittekamp v. Gulf & W., Inc., 991 F.2d 1137, 1141 (3d Cir. 1993).

No witnesses suggested that what Jones did was "extraordinary"; Jones and Hill for the plaintiff, and Vagn Ejsing and Curran for the defense, all testified only that it was common for longshoremen to strike tight wing nuts to loosen them. Curran did give testimony critical of Jones, but at best it can be taken to indicate only negligence, not extraordinariness.15 Curran was asked on direct "what if anything [Jones] should [] have done as a prudent longshoreman," and he replied: "He should have called the supervisor." App. 365. But third-party negligence does not wipe out defendants' liability in maritime law. If we allowed the instruction on these facts, then regardless of labels, we risk allowing that result.

We assume, though we do not decide, that a reasonable jury could have found that Jones (and thus the stevedore through respondeat superior) was negligent. If, however, the superseding cause instruction was interpreted by the jury to mean that the shipowner was thereby relieved of liability as a matter of law, the result would be directly contrary to the LHWCA and Edmonds. Because of the danger that ordinary stevedore negligence might be inferred to cut off shipowner liability entirely, courts must be wary of giving superseding cause instructions, and should do so, if at all, only when there is an adequate evidentiary basis. In this case, there was none.

The District Court correctly instructed the jury that superseding cause is a defense to a negligence claim that could otherwise be made out: "And bear in mind that although generally in this case, the plaintiff has the burden of proof by a preponderance of the evidence, in terms of this superseding cause, the defendant has the burden of proof by a preponderance of the evidence." App. 442. To support the superseding cause instruction, accordingly, the defendants needed to have put on at least some evidence of all of the following: (1) they did not in fact know that longshoremen hit the wing nuts with wrenches, (2) they had no reason to know that longshoremen might hit the wing nuts with wrenches, (3) it was highly extraordinary for longshoremen to hit the wing with wrenches, and (4) Jones' hitting the wing nut with the wrench was not the consequence of any act or failure to act by the shipowners. See App. 442 (instruction); Restatement (Second) § 447. The record, as described above, simply cannot support any of these propositions.16

Trial courts must carefully circumscribe their instructions to the jury, because while the jury is the factfinder, the range of facts capable of being found in a given case is a question of law. We will vacate a jury verdict if, "viewing the evidence in the light most favorable to the nonmovant and giving it the advantage of every fair and reasonable inference, there is insufficient evidence from which a jury reasonably" could have reached the result it did. Lightning Lube v. Witco Corp., 4 F.3d 1153, 1166 (3d Cir. 1993). Sofec's holding that superseding cause is a viable maritime category does not relieve the courts of our responsibility to ensure that the instruction is only given where the record evidence is sufficient to reasonably support a superseding cause finding. The evidence here could not support the finding, so the charge should not have been given.

We do not hold that on this record a properly instructed jury could not reasonably have found that the ship was not negligent. Rather, we hold that such a finding cannot legally have been grounded in superseding cause in this case. A superseding cause instruction will be permissible on remand only if evidence is presented showing that Jones's actions were outside the normal range of customary longshoremen's practices. Because there was no evidentiary basis for a superseding cause finding, the District Court erred in giving a superseding cause instruction.

The final two issues before us are the District Court's admission of certain testimony of the defense's expert over Hill's objection of unfair surprise, and its denial of Hill's request for a res ipsa loquitur instruction. We find no error in either ruling.

Hill contends that he was unfairly surprised when the defendant's expert, Walter Curran, testified that the accident could not have happened as Jones described it in his testimony. Defense counsel asked Curran to opine on whether, if Jones had struck a frozen wing nut with a wrench, the lashing rod could have sprung out and flown across the hold. App. 370. Curran responded that such a scenario was "physically impossible," and that in his opinion, in order for a lashing rod to spring out of its housing, the turnbuckle would have to be partially loosened before being struck with the wrench. App. 373-74. Curran therefore hypothesized that Jones had partially loosened the turnbuckle, but had attempted to remove the lashing rod before the turnbuckle was fully loose, causing the bar to flex "like a spring." "When you give it a whack under those circumstances," Curran testified, "it could very well spring free." Hill argues that that opinion — that Jones must have partially loosened the bar before striking the bolt — was not in Curran's report.

Curran's report states his opinion "that the turnbuckle in question was not `frozen' due to rust as alleged." App. 461. The report then proposes an alternative scenario in which turnbuckles can become frozen — the very alternative scenario Curran described at trial: "[I]f the longshoremen attempt to loosen the turnbuckle without first backing off the wing nut the turnbuckle can get `bound up.' This is a self-inflicted situation borne of carelessness on the part of the longshoremen." App. 461. The report thus says what Curran said at trial: that the accident was likely precipitated by Jones's failure to fully loosen the turnbuckle before attempting to remove the lashing rod.

The report, to be sure, does not contain the other opinion Curran offered at trial: that it is "physically impossible" for a lashing rod to spring free without being first partially loosened. To that extent, then, Curran's trial testimony exceeded the scope of his report. But the permissible scope of expert testimony is quite broad, and District Courts are vested with broad discretion in making admissibility determinations. The District Court notes that "Plaintiffs do not cite, and the Court is not aware of, any bright line rule that every opinion by an expert must be preliminarily stated in the report, or forever be precluded." App. 24. This Court is similarly unaware of such a rule. Cf. Weinstein's Federal Evidence § 403.02[4][a] (2d ed. 2005) ("Some members of the bar would have preferred to add surprise to Rule 403, but the Advisory Committee rejected this suggestion. The modern shift in attitude does not require the recognition of surprise, standing alone, as a ground for exclusion.") (internal citation omitted).

Furthermore, Curran's testimony was elicited in rebuttal of Jones's account of the accident. Curran was asked his opinion of the likelihood of the sequence of events Jones had described in his testimony. Such a rebuttal is thoroughly in accord with Federal Rule of Evidence 703, which provides that "[t]he facts or data in the particular case upon which an expert bases an opinion or inference may be those perceived by or made known to the expert at or before the hearing." The Advisory Committee's Note on the rule states that the rule intends to "reflect[] [the] existing practice . . . [of] having the expert attend the trial and hear the testimony establishing the facts." We can discern no error in admitting expert opinion offered in response to prior trial testimony. Furthermore, whatever surprise there might have been was adequately cured by Hill's extensive cross-examination of Curran. And if cross-examination was insufficient, then rather than resting on an objection, a better procedure would be to request a sidebar on the issue of surprise, and even a recess to investigate the new evidence. See Johnson v. H.K. Webster, Inc., 775 F.2d 1, 8 (1st Cir. 1985) ("Courts have looked with disfavor upon parties who claim surprise and prejudice but who do not ask for a recess so they may attempt to counter the opponent's testimony."). Where surprise has occurred, the appropriate action remains within the trial court's discretion, informed by the following factors: "(1) the prejudice or surprise in fact to the opposing party, (2) the ability of the party to cure the prejudice, (3) the extent of disruption of the orderly and efficient trial of the case, and (4) the bad faith or willfulness of the non-compliance." Hurley v. Atlantic City Police Dept., 174 F.3d 95, 113 (3d Cir. 1999) (citations and quotations omitted).

Finally, we can find no error in the District Court's denial of Hill's request for a res ipsa loquitur instruction. A res ipsa loquitur instruction will be justified if the plaintiff can show, inter alia, that "other responsible causes, including the conduct of the plaintiff and third persons, are sufficiently eliminated by the evidence." Restatement (Second) of Torts § 328. The parties in this case agree that the lashing rod sprang free when Jones hit the wing nut with his wrench. They thus agree that Jones' act was a proximate cause of the accident. It is therefore a question of fact whether Jones's act could have been the cause of the accident even without any negligence by the ship. The District Court was therefore correct not to charge res ipsa loquitur.17

Having determined that the jury instructions given by the District Court were erroneous in the respects explained above, and that Hill was prejudiced by these errors, we will vacate the District Court's order, and remand for a new trial.

RENDELL, Circuit Judge.

I agree that this case should be returned to the District Court for a new trial. But I do so on a very different basis, namely because the defendants' expert's opinion at trial that the accident could not have happened as Jones testified, and was "physically impossible," was radically different from the opinion that he rendered in his pre-trial report. I do not agree that the superseding cause instruction forms an appropriate basis for reversal.

I disagree with the majority's exploration of the theoretical difficulty with the use of superseding cause instructions in a comparative negligence scheme as misplaced and unnecessary. It is misplaced because the use of such instructions has been criticized where they can operate to absolve a defendant from liability based on the plaintiff's own negligence, and thus re-introduce principles of contributory negligence into a comparative fault system. In such situations, the application of the superseding cause doctrine arguably gives defendants a windfall where they would have been liable, at least in part, on the basis of comparative fault. Here there is no allegation that Hill was negligent; the only issue is the extent of the negligence of a third party, Jones, and its effect on the negligence and liability of the shipowner defendants. I suggest that the concept of "heightened vigilance" in giving superseding cause instructions that the majority announces has no place in the fact pattern before us. Rather, it should be reserved for those situations where the instruction would operate like contributory negligence, inequitably denying plaintiff any recovery on the basis of his own negligence. Since that is not at issue here, the concern expressed, and rule announced, by the majority is like apples to the oranges before us and should be left for another day and another case.18

Furthermore, the entire discussion of superseding cause is unnecessary because the jury found that neither defendant was negligent. The majority assumes that the jury might have been saying on the verdict slip, when marking "no" to the questions "Was Defendant Schiffahrtsgesellschaft MS Priwall mbH & Co. KG, the owner of the MS Sea Panther, negligent?" and "Was Defendant Reederei F. Laiesz G.m.b.h., Rostock, the operator of the MS Sea Panther, negligent?" that the defendants were negligent but that their negligence was not the cause of Hill's injuries. I find this uncalled for, as we presume that juries follow instructions, see Citizens Fin. Group, Inc. v. Citizens Nat. Bank of Evans City, 383 F.3d 110, 133 (3d Cir. 2004), and the instructions given by the District Court made it clear that superseding cause operates to absolve an otherwise negligent defendant from liability, not to require a finding of no negligence before the issue of causation is addressed.19

In addition, in light of the second question on the verdict form, we need not speculate that the jury might have found the defendants not negligent when what they really meant was that their negligence was not a legal cause of Hill's injury. That question asked specifically whether, if one or both of the defendants was negligent, "that negligence, in whole or in part, cause[d] any injury and damage to Plaintiff Cornelius Hill." Had the jury really concluded, based on a theory of superseding cause, that the defendants were negligent but that their negligence was not the cause of Hill's injury, the jury form provided it with a way of saying so. Because it did not, we should assume that the jury's verdict means what it says—that the jury found that neither of the defendants was negligent. The issue of the propriety of a causation instruction is not relevant, as any error in giving it was harmless.

My difficulty with allowing Mr. Curran's opinion given at trial is that it clearly surprised the plaintiffs and materially prejudiced them with respect to the crucial question in the case in a way that was impossible to cure through cross examination. Although a district court retains discretion to determine whether expert testimony should be excluded when it exceeds the scope of the pretrial report, its discretion is not unfettered. We apply a multi-factor test in determining whether a district court has abused this discretion, considering (1) the prejudice or surprise to the party against whom the witness testifies; (2) the ability of the opposing party to cure any such prejudice; (3) the effect of allowing or excluding testimony on the trial of the case; (4) any bad faith or wilfulness involved in the presentation of such testimony; and (5) the importance of the testimony to be excluded or omitted to the overall case. Quinn v. Consol. Freightways Corp., 283 F.3d 572, 577 (3d Cir. 2002) (citing Meyers v. Pennypack Woods Home Ownership Ass'n, 559 F.2d 894, 904-05 (3d Cir. 1977)).

The expert opinion given at trial altered the defendant's theory of the case radically. The expert's pre-trial opinion was that Mr. Hill's injury was caused not by the frozen turnbuckle, but by Mr. Jones's negligence in:

1. failing to notify his supervisor of the frozen turnbuckle;

2. failing to request assistance in freeing the frozen turnbuckle;

3. striking the turnbuckle with his wrench; and

4. failing to warn Mr. Hill of his intended action.

However, at trial, Curran characterized Jones's testimony that the lashing rod came out of its casing when he hit the frozen turnbuckle with his wrench as "physically impossible." He opined that hitting a frozen turnbuckle would not cause anything to "fly out or budge," and that the accident could only have happened once Jones managed to loosen the turnbuckle, when Jones either failed to "hold the bar the way he should have done," or whacked the loosened lashing rod with his wrench, so that it "act[ed] like a spring and . . . spr[u]ng free with the bottom coming out first." Although the expert explained his new opinion away as just filling in the "details" of his report, it represented an entirely new and, from the plaintiff's perspective, unanticipated explanation of the events leading up to Hill's injuries.

The cross examination by plaintiff's counsel, thought by the District Court to be adequate, could only attack the difference between the defense expert's testimony and his pre-trial report, not the substance of that testimony. Any attempt to undercut its substance would only have reinforced the expert's view that the plaintiff's theory of how the accident occurred was "physically impossible." Surely the jury could not appreciate how problematic this situation was for plaintiff's counsel, caught unawares by an expert's opinion that totally obliterated the key assertion of his case, namely, that the turnbuckle's rusted and frozen condition caused the plaintiff's injury.

The District Court erred not only in not seeing the drastic difference between these two opinions, but also in justifying the introduction of a new opinion based on the fact that Jones's testimony at trial was more comprehensive than at his deposition. But the fault for failing to adequately depose Jones lay with the defendants themselves. The proper approach for the defendants in that situation was not to elicit a new expert opinion without prior notice, but, rather, to ask for time to submit a supplemental opinion based on the additional facts adduced at trial.

For the foregoing reasons, I conclude that the District Court abused its discretion in allowing the defense expert to offer an opinion that was outside the scope of his pre-trial report. Accordingly, I concur in the majority's decision to reverse the judgment below and remand the case for a new trial.