In Liriano v. Hobart Corp., 132 F.3d 124 (2d Cir.1998) ("Liriano I "), we certified to the New York Court of Appeals the question of whether a manufacturer can be liable under a failure-to-warn theory in a case in which the substantial modification defense would preclude liability under a design defect theory. See id. at 132. We also certified the question of whether, if failure-to-warn liability could exist, it would nonetheless be unavailable as a matter of law on the facts of the present case. See id. The New York Court of Appeals answered the first question in the affirmative and declined to answer the second. See Liriano v. Hobart Corp., 92 N.Y.2d 232, 236, 700 N.E.2d 303, 304, 677 N.Y.S.2d 764, 765 (1998) ("Liriano II "). Consequently, we now address the second question ourselves, and we find it to be a close one. Viewing the facts, as we must, in the light most favorable to the plaintiff, we resolve that question in the negative. We also find that all other claims the appellants have raised on appeal lack merit. We therefore affirm the decision of the district court granting judgment and damages for the plaintiff.

The facts of this case are set out in Liriano I and again in Liriano II. See Liriano I, 132 F.3d at 125-26; Liriano II, 92 N.Y.2d at 236-37, 700 N.E.2d at 305, 677 N.Y.S.2d at 766. We repeat only those that are necessary to resolve the issues that remain before us.

Luis Liriano was severely injured on the job in 1993 when his hand was caught in a meat grinder manufactured by Hobart Corporation ("Hobart") and owned by his employer, Super Associated ("Super"). The meat grinder had been sold to Super with a safety guard, but the safety guard was removed while the machine was in Super's possession and was not affixed to the meat grinder at the time of the accident. The machine bore no warning indicating that the grinder should be operated only with a safety guard attached.

Liriano sued Hobart under several theories, including failure to warn. Hobart brought a third-party claim against Super. The United States District Court for the Southern District of New York (Shira A. Scheindlin, J.) dismissed all of Liriano's claims except the one based on failure to warn, and the jury returned a verdict for Liriano on that claim. It attributed five percent of the liability to Hobart and ninety-five percent to Super. The district court then held a partial retrial limited to the issue of whether and to what extent Liriano was responsible for his own injury. On that retrial, the jury assigned Liriano one-third of the fault.

Before she entered judgment based on these calculations, Judge Scheindlin granted Liriano's motion to add $21,252.34 to the total damage award. The increase reflected the amount of a hospital bill that had been submitted to the jury during trial but somehow had not been included in the initial damage calculation.

Hobart and Super appealed, arguing (1) that as a matter of law, there was no duty to warn, and (2) that even if there had been a duty to warn, the evidence presented was not sufficient to allow the failure-to-warn claim to reach the jury. Super further argued (3) that the district court erred in holding a retrial on the issue of Liriano's comparative fault without also retrying Hobart's and Super's shares of fault relative to each other, and (4) that the district court's addition of the hospital bill to its damage calculation was an impermissible additur. We certified questions (1) and (2) to the New York Court of Appeals. See Liriano I, 132 F.3d at 132. That Court answered question (1) in Liriano's favor, saying that there can indeed be a duty to warn in a case like this one. See Liriano II, 92 N.Y.2d at 243, 700 N.E.2d at 309, 677 N.Y.S.2d at 770. The Court of Appeals, however, declined to answer question (2). See id.

We now resolve questions (2), (3), and (4).

Hobart makes two arguments challenging the sufficiency of the evidence. The first concerns the obviousness of the danger that Liriano faced, and the second impugns the causal relationship between Hobart's negligence and Liriano's injury. Each of these arguments implicates issues long debated in the law of torts. With respect to the asserted clarity of the danger, the question is when a danger is so obvious that a court can determine, as a matter of law, that no additional warning is required. With respect to causation, the issue is whether a jury may infer that a defendant's particular negligence was the cause-in-fact of a plaintiff's actual injury from the general fact that negligence like the defendant's tends to cause injuries like the plaintiff's. The obviousness question was the subject of an important but now generally rejected opinion by Justice Holmes, then on the Massachusetts Supreme Judicial Court;1 the causation question is answered in a celebrated opinion of Judge Cardozo, then on the New York Court of Appeals.2 We examine each in turn.

(1) Obviousness

More than a hundred years ago, a Boston woman named Maria Wirth profited from an argument about obviousness as a matter of law that is very similar to the one Hobart urges today. See Lorenzo v. Wirth, 170 Mass. 596, 49 N.E. 1010 (1898). Wirth was the owner of a house on whose property there was a coal hole. See id. at 599, 49 N.E. at 1010. The hole abutted the street in front of the house, and casual observers would have no way of knowing that the area around the hole was not part of the public thoroughfare. See id. at 599, 49 N.E. at 1011. A pedestrian called Lorenzo fell into the coal hole and sued for her injuries. See id. Writing for a majority of the Supreme Judicial Court of Massachusetts, Oliver Wendell Holmes, Jr., held for the defendant. He noted that, at the time of the accident, there had been a heap of coal on the street next to the coal hole, and he argued that such a pile provided sufficient warning to passers-by that they were in the presence of an open hole. "A heap of coal on a sidewalk in Boston is an indication, according to common experience, that there very possibly may be a coal hole to receive it." Id. at 601, 49 N.E. at 1011. And that was that.

It was true, Holmes acknowledged, that "blind men, and foreigners unused to our ways, have a right to walk in the streets," id. at 600, 49 N.E. at 1011, and that such people might not benefit from the warning that piles of coal provided to sighted Bostonians. But Holmes wrote that coal-hole cases were simple, common, and likely to be oft repeated, and he believed it would be better to establish a clear rule than to invite fact-specific inquiries in every such case. "In simple cases of this sort," he explained, "courts have felt able to determine what, in every case, however complex, defendants are bound at their peril to know." Id. With the facts so limited, this was an uncomplicated case in which the defendant could, as a matter of law, rely on the plaintiff's responsibility to know what danger she faced.

Justice Knowlton disagreed. See id. at 601-04, 49 N.E. at 1011-13 (Knowlton, J., dissenting). His opinion delved farther into the particular circumstances than did Holmes's opinion for the majority. In so doing, he showed that Lorenzo's failure to appreciate her peril might have been foreseen by Wirth and hence that Wirth's failure to warn might constitute negligence. He noted, for example, that the accident occurred after nightfall, when Lorenzo perhaps could not see, or recognize, the heap of coal for what it was. See id. at 603, 49 N.E. at 1012. There was "a throng of persons" on the street, such that it would have been difficult even in daylight to see very far ahead of where one was walking. See id. at 603-04, 49 N.E. at 1012. And the plaintiff was, in fact, a foreigner unused to Boston's ways. "[S]he had just come from Spain, and had never seen coal put into a cellar through a coal hole." Id. at 603, 49 N.E. at 1012. In sum, the case was not the "simple" one that Holmes had made it out to be. What is more, none of the facts he recited was either unusual or unforeseeable by Wirth. "What kind of conduct is required under complex conditions, to reach the usual standard of due care, namely, the ordinary care of persons of common prudence, is a question of fact .... [and thus] a question for a jury." Id. at 604, 49 N.E. at 1012-13 (emphasis added). Even cases involving "obvious" dangers like coal holes, Knowlton believed, might not be resolvable as matters of law when viewed in the fullness of circumstances that rendered the issue less clear than it would be when posed in the abstract.

Holmes commanded the majority of the Supreme Judicial Court in 1898, but Knowlton's position has prevailed in the court of legal history. " '[T]he so-called Holmes view--that standards of conduct ought increasingly to be fixed by the court for the sake of certainty--has been largely rejected.... The tendency has been away from fixed standards and towards enlarging the sphere of the jury.' " Fowler V. Harper, Fleming James, Jr., & Oscar S. Gray, The Law of Torts § 15.3, at 358-59 n. 16 (2d ed.1986) (hereinafter Harper & James) (quoting Nuckoles v. F.W. Woolworth Co., 372 F.2d 286, 289 (4th Cir.1967)).

The courts of New York have several times endorsed Knowlton's approach and ruled that judges should be very wary of taking the issue of liability away from juries, even in situations where the relevant dangers might seem obvious, and especially when the cases in question turn on particularized facts. See, e.g., Havas v. Victory Paper Stock Co., 49 N.Y.2d 381, 388, 402 N.E.2d 1136, 1139-40, 426 N.Y.S.2d 233, 237 (1980) (stating that negligence is an issue "particularly appropriate to leave ... to the jury" because of the "idiosyncratic nature of most tort cases" and because "reasonableness" is generally appropriate for jury resolution); Cabri v. Long Island R.R. Co., 306 N.Y. 765, 118 N.E.2d 475 (1954) (holding that the danger of crossing railroad tracks is not so obvious as to prevent the issue of contributory negligence from reaching the jury).3 See also Stagl v. Delta Airlines, Inc., 52 F.3d 463, 470-71 (2d Cir.1995) (noting the demise of the Holmes position and collecting cases); Fleming James, Jr., Geoffrey C. Hazard, Jr., & John Leubsdorf, Civil Procedure § 7.21, at 371 (4th ed.1992) (stating that the decline of the Holmes view has been especially pronounced in tort suits for personal injury); Harper & James, § 17.2, at 571-72 nn. 8-13, 573-74 nn. 22-26, 576-78 nn. 33-37 (collecting many authorities).

But the secular decline of the Holmes position and the concomitant tendency of the New York Court of Appeals to permit issues of obviousness to go to the jury do not fully dispose of the question before us. After all, as Holmes himself might have cautioned, general trends are far from conclusive in concrete cases. Cf. Lochner v. New York, 198 U.S. 45, 75, 25 S. Ct. 539, 49 L. Ed. 937 (1905) (Holmes, J., dissenting). And it is not surprising that there have been situations in which New York state courts have deemed dangers to be sufficiently clear so that warnings were, as a matter of law, not necessary. See, e.g., Dickerson v. George J. Meyer Mfg., 248 A.D.2d 970, 971, 669 N.Y.S.2d 1001, 1002 (4th Dep't 1998) (holding that there is no duty to warn of the danger of closely examining the mechanical workings of a machine while the machine is operating); Pigliavento v. Tyler Equip. Corp., 248 A.D.2d 840, 842, 669 N.Y.S.2d 747, 749 (3d Dep't 1998) (holding that there is no duty to warn of the danger of falling from an unguarded platform on a concrete mixer truck); Caris v. Mele, 134 A.D.2d 475, 476, 521 N.Y.S.2d 260, 261 (2d Dep't 1987) (holding that there is no duty to warn of the danger of diving headfirst into an above-ground swimming pool only four feet deep).

If the question before us were, therefore, simply whether meat grinders are sufficiently known to be dangerous so that manufacturers would be justified in believing that further warnings were not needed, we might be in doubt. On one hand, just as a coal hole was deemed a danger appreciated by most Bostonians in 1898, so most New Yorkers would probably appreciate the danger of meat grinders a century later. Any additional warning might seem superfluous. On the other hand, Liriano was only seventeen years old at the time of his injury and had only recently immigrated to the United States. He had been on the job at Super for only one week. He had never been given instructions about how to use the meat grinder, and he had used the meat grinder only two or three times. And, as Judge Scheindlin noted, the mechanism that injured Liriano would not have been visible to someone who was operating the grinder. It could be argued that such a combination of facts was not so unlikely that a court should say, as a matter of law, that the defendant could not have foreseen them or, if aware of them, need not have guarded against them by issuing a warning. That argument would draw strength from the Court of Appeals' direction that the question of whether a warning was needed must be asked in terms of the information available to the injured party rather than the injured party's employer, see Liriano II, 92 N.Y.2d at 241, 700 N.E.2d at 308, 677 N.Y.S.2d at 769, and its added comment that "in cases where reasonable minds might disagree as to the extent of the plaintiff's knowledge of the hazard, the question is one for the jury." Id.

Nevertheless, it remains the fact that meat grinders are widely known to be dangerous. Given that the position of the New York courts on the specific question before us is anything but obvious, we might well be of two minds as to whether a failure to warn that meat grinders are dangerous would be enough to raise a jury issue.4

But to state the issue that way would be to misunderstand the complex functions of warnings. As two distinguished torts scholars have pointed out, a warning can do more than exhort its audience to be careful. It can also affect what activities the people warned choose to engage in. See James A. Henderson, Jr., and Aaron D. Twerski, Doctrinal Collapse in Products Liability: The Empty Shell of Failure to Warn, 65 N.Y.U. L.Rev. 265, 285 (1990). And where the function of a warning is to assist the reader in making choices, the value of the warning can lie as much in making known the existence of alternatives as in communicating the fact that a particular choice is dangerous. It follows that the duty to warn is not necessarily obviated merely because a danger is clear.

To be more concrete, a warning can convey at least two types of messages. One states that a particular place, object, or activity is dangerous. Another explains that people need not risk the danger posed by such a place, object, or activity in order to achieve the purpose for which they might have taken that risk. Thus, a highway sign that says "Danger--Steep Grade" says less than a sign that says "Steep Grade Ahead--Follow Suggested Detour to Avoid Dangerous Areas."

If the hills or mountains responsible for the steep grade are plainly visible, the first sign merely states what a reasonable person would know without having to be warned. The second sign tells drivers what they might not have otherwise known: that there is another road that is flatter and less hazardous. A driver who believes the road through the mountainous area to be the only way to reach her destination might well choose to drive on that road despite the steep grades, but a driver who knows herself to have an alternative might not, even though her understanding of the risks posed by the steep grade is exactly the same as those of the first driver. Accordingly, a certain level of obviousness as to the grade of a road might, in principle, eliminate the reason for posting a sign of the first variety. But no matter how patently steep the road, the second kind of sign might still have a beneficial effect. As a result, the duty to post a sign of the second variety may persist even when the danger of the road is obvious and a sign of the first type would not be warranted.

One who grinds meat, like one who drives on a steep road, can benefit not only from being told that his activity is dangerous but from being told of a safer way. As we have said, one can argue about whether the risk involved in grinding meat is sufficiently obvious that a responsible person would fail to warn of that risk, believing reasonably that it would convey no helpful information. But if it is also the case--as it is--that the risk posed by meat grinders can feasibly be reduced by attaching a safety guard, we have a different question. Given that attaching guards is feasible, does reasonable care require that meat workers be informed that they need not accept the risks of using unguarded grinders? Even if most ordinary users may--as a matter of law--know of the risk of using a guardless meat grinder, it does not follow that a sufficient number of them will--as a matter of law--also know that protective guards are available, that using them is a realistic possibility, and that they may ask that such guards be used. It is precisely these last pieces of information that a reasonable manufacturer may have a duty to convey even if the danger of using a grinder were itself deemed obvious.

Consequently, the instant case does not require us to decide the difficult question of whether New York would consider the risk posed by meat grinders to be obvious as a matter of law. A jury could reasonably find that there exist people who are employed as meat grinders and who do not know (a) that it is feasible to reduce the risk with safety guards, (b) that such guards are made available with the grinders, and (c) that the grinders should be used only with the guards. Moreover, a jury can also reasonably find that there are enough such people, and that warning them is sufficiently inexpensive, that a reasonable manufacturer would inform them that safety guards exist and that the grinder is meant to be used only with such guards. Thus, even if New York would consider the danger of meat grinders to be obvious as a matter of law, that obviousness does not substitute for the warning that a jury could, and indeed did, find that Hobart had a duty to provide. It follows that we cannot say, as a matter of law, that Hobart had no duty to warn Liriano in the present case. We therefore decline to adopt appellants' argument that the issue of negligence was for the court only and that the jury was not entitled, on the evidence, to return a verdict for Liriano.

(2) Causation

On rebriefing following the Court of Appeals decision, Hobart has made another argument as to why the jury should not have been allowed to find for the plaintiff. In this argument, Hobart raises the issue of causation. It maintains that Liriano "failed to present any evidence that Hobart's failure to place a warning [on the machine] was causally related to his injury." Whether or not there had been a warning, Hobart says, Liriano might well have operated the machine as he did and suffered the injuries that he suffered. Liriano introduced no evidence, Hobart notes, suggesting either that he would have refused to grind meat had the machine borne a warning or that a warning would have persuaded Super not to direct its employees to use the grinder without the safety attachment.

Hobart's argument about causation follows logically from the notion that its duty to warn in this case merely required Hobart to inform Liriano that a guard was available and that he should not use an unguarded grinder. The contention is tightly reasoned, but it rests on a false premise. It assumes that the burden was on Liriano to introduce additional evidence showing that the failure to warn was a but-for cause of his injury, even after he had shown that Hobart's wrong greatly increased the likelihood of the harm that occurred. But Liriano does not bear that burden. When a defendant's negligent act is deemed wrongful precisely because it has a strong propensity to cause the type of injury that ensued, that very causal tendency is evidence enough to establish a prima facie case of cause-in-fact. The burden then shifts to the defendant to come forward with evidence that its negligence was not such a but-for cause.

We know, as a general matter, that the kind of negligence that the jury attributed to the defendant tends to cause exactly the kind of injury that the plaintiff suffered. Indeed, that is what the jury must have found when it ruled that Hobart's failure to warn constituted negligence. In such situations, rather than requiring the plaintiff to bring in more evidence to demonstrate that his case is of the ordinary kind, the law presumes normality and requires the defendant to adduce evidence that the case is an exception. Accordingly, in a case like this, it is up to the defendant to bring in evidence tending to rebut the strong inference, arising from the accident, that the defendant's negligence was in fact a but-for cause of the plaintiff's injury. See Zuchowicz v. United States, 140 F.3d 381, 388 nn. 6-7, 390-91 (2d Cir.1998).

This shifting of the onus procedendi has long been established in New York. Its classic statement was made more than seventy years ago, when the Court of Appeals decided a case in which a car collided with a buggy driving after sundown without lights. See Martin v. Herzog, 228 N.Y. 164, 170, 126 N.E. 814, 816 (1920). The driver of the buggy argued that his negligence in driving without lights had not been shown to be the cause-in-fact of the accident. Writing for the Court, Judge Cardozo reasoned that the legislature deemed driving without lights after sundown to be negligent precisely because not using lights tended to cause accidents of the sort that had occurred in the case. See id. at 168, 126 N.E. at 815. The simple fact of an accident under those conditions, he said, was enough to support the inference of but-for causal connection between the negligence and the particular accident. See id. at 170, 126 N.E. at 816. The inference, he noted, could be rebutted. But it was up to the negligent party to produce the evidence supporting such a rebuttal. See id.

The words that Judge Cardozo applied to the buggy's failure to use lights are equally applicable to Hobart's failure to warn: "If nothing else is shown to break the connection, we have a case, prima facie sufficient, of negligence contributing to the result." Id. Under that approach, the fact that Liriano did not introduce detailed evidence of but-for causal connection between Hobart's failure to warn and his injury cannot bar his claim. His prima facie case arose from the strong causal linkage between Hobart's negligence and the harm that occurred. See Guido Calabresi, Concerning Cause and the Law of Torts: An Essay for Harry Kalven, Jr., 43 U. Chi. L.Rev. 69 (1975) (describing the concept of "causal link"). And, since the prima facie case was not rebutted, it suffices.5

The district court ordered a retrial on the issue of Liriano's comparative negligence but declined to do the same as to the proper allocation of fault between the two defendants. As a general matter, judges may order retrial on some but not all issues in a case, and they may do so even when the issues are related. Under Akermanis v. Sea-Land Service, Inc., 688 F.2d 898 (2d Cir.1982), a district judge has discretion to order retrial of the issue of a plaintiff's comparative negligence without requiring the fact-finder to reconsider the total extent of damages. See id. at 906-07. As we explained in Crane v. Consolidated Rail Corp., 731 F.2d 1042 (2d Cir.1984), Akermanis teaches that a judge need not infer jury error on a given issue merely because she finds jury error on a related issue. See id. at 1050-51.

It is possible that a judge might conclude that the facts of a given case required reexamining the relative liabilities of two defendants as part of a retrial on the issue of a plaintiff's comparative negligence. But on the facts of this case, and in light of Akermanis and Crane, such a retrial is not mandated. The district court decided that the relative shares of fault attributed to Hobart and Super were "amply supported by the record." We affirm that decision.

Appellant Super argues that the district court granted an impermissible additur when it increased the damage award to account for a hospital bill whose amount and nature were not in dispute. Additur is a practice by which a judge offers a defendant the choice between facing a retrial and accepting a damage award higher than that determined by the jury. Like remittitur, it is a tool that judges use to fix damages--something that can generally be done only by the fact-finder--without actually having to hold a second trial.

Under the rule of Dimick v. Schiedt, 293 U.S. 474, 55 S. Ct. 296, 79 L. Ed. 603 (1935), federal courts are denied the same freedom to use additur that is enjoyed by many state court judges. In this case, however, there was no true additur. The district court did not divine a figure and then make the defendants choose between an increased damage award and a new trial. It simply adjusted the jury award to account for a discrete item that manifestly should have been part of the damage calculations and as to whose amount there was no dispute. When a jury has already found liability, federal courts may make such adjustments without running afoul of Dimick. See, e.g., United States Equal Employment Opportunity Comm'n v. Massey Yardley Chrysler Plymouth, Inc., 117 F.3d 1244, 1252-53 (11th Cir.1997); Decato v. Travelers Ins. Co., 379 F.2d 796, 798 (1st Cir.1967).

The district court did not err. We affirm its decision in all respects.

When this diversity appeal was first before us, we were sufficiently unsure about how the case should be decided under New York law that we asked the New York Court of Appeals to answer two questions. The first was whether a manufacturer that designs a product in a reasonably safe manner can nonetheless be held negligent for failure to warn about a product's danger that results from a modification of the product after its sale. The New York Court of Appeals answered that such failure-to-warn liability can exist. In order to understand how New York would apply its tort law in this rather unusual context, we also asked the State's highest court whether such failure-to-warn liability is a fact issue for a jury under the circumstances of this case. Unfortunately, the Court of Appeals declined to answer the second question.

The Court of Appeals' decision not to answer our fact-specific inquiry is understandable, since the certification process is most usefully employed to resolve unsettled issues of law. However, in many contexts a full understanding of a state's legal standard may be achieved only by seeing how a state's highest court would apply that standard to the specific facts of a case. I think that is so in this case, and I would have been extremely interested to see how the New York Court of Appeals would have applied its failure-to-warn standard to the facts of Liriano's injury resulting from his somehow placing his hand into the spout of Hobart's meat-grinder from which the safety guard had been removed long after Hobart sold the product. Nevertheless, in the absence of a definitive Court of Appeals answer to our second question, we, as a diversity court, must make our best prediction as to how New York courts would answer that question had this case been tried in a state court. I am in considerable doubt as to what that prediction should be, but for reasons similar to those set forth by the majority, I agree that the New York Court of Appeals would most likely permit the liability issue in this case to be decided by a jury, and I therefore concur.

Like the majority, I see no need to decide how this case would have been resolved had Hobart manufactured its meat-grinder without a safety guard--the circular plate above the spout with holes large enough for pieces of meat but small enough to prevent insertion of a hand. Instead, the issue for us is whether a New York court would permit a jury to consider liability for failure to warn of the danger of using the meat-grinder after a safety guard, originally installed, has been removed. Some might think it should make no difference whether a safety guard was originally installed so long as the machine the user confronts lacks such a guard. After all, the obviousness of the danger of placing one's hand into the unguarded spout of a meat-grinder is the same for a person using a machine that was never equipped with a guard as for a person using a machine manufactured with a guard that was later removed. But the circumstances confronting the user of these two machines vary in two respects potentially relevant to the issue in this case.

First, if the second machine visibly indicates that it once was equipped with a safety guard, the user is alerted to the danger of using the machine without a device that the manufacturer thought was advisable for enhancing safety. For example, if the safety guard was attached by some form of latch, the opening of which permitted removal of the guard, and if the latch bore the legend "lock safety guard here," the user of a machine from which the guard had been unlatched and removed could reasonably be found by a jury to have had his appreciation of the danger enhanced, beyond what he should reasonably have perceived simply by seeing the open spout and knowing what happens to objects placed into that spout. To whatever extent such enhanced appreciation of the risk is relevant, it undermines the plaintiff's claim, since a manufacturer need not warn of an obvious risk and certainly need not do so as to a risk that is even more obvious than the risk presented by a machine that was never equipped with a guard. The tendency of a removed guard to enhance the user's awareness of the risk is uncertain in this case, however, because the safety guard that was originally installed was not secured by a latch (and there was no legend concerning its attachment). Instead, is was bolted to the meat-grinder, and the record is unclear as to what conclusion, if any, a user could reasonably draw from the appearance of the machine as to the previous existence of a guard.

The second circumstance implicated by the previous existence of a safety guard is the one discussed in the Court's opinion--the availability of the option of insisting upon a machine that has a guard, rather than facing only the choice between using a machine with a guard and not using the machine that lacks a guard. The Court's opinion offers the example of a steep road marked in one instance with a sign that warns "Danger--Steep Grade" and in another instance with a more informative sign indicating the option of an alternate route that avoids the steep grade. The example is not precisely analogous to our case because the record gives no indication that an alternate machine with a safety guard was as readily available as an alternate driving route. Moreover, the option of a driver to choose an indicated alternate driving route immediately available is more realistic than the option of a supermarket employee to insist on a machine the availability of which is entirely uncertain. Nevertheless, the Court's analogy usefully indicates a circumstance common to both the Court's example and our case: the alternate means of proceeding more safely is an option known to the entity that bears a relationship to the dangerous condition and is not known to the person encountering the danger. When that disparity of knowledge exists, may liability be imposed by a fact-finder for failure to warn of the alternative?

This becomes the critical question in this case, and we have no firm basis on which to predict the answer New York's highest court would give to it. An injury occurring as a result of a rather obvious danger but one that might have been avoided by an alternative known to a product manufacturer and not known to a likely user of the product has not been considered by the New York Court of Appeals, and appears not to have been directly confronted by other New York courts. Perhaps the most relevant clue is the observation that the Court of Appeals made in its response to our first question in this case: "where reasonable minds might disagree as to the extent of the plaintiff's knowledge of the hazard, the question is one for the jury." Liriano v. Hobart Corp., 92 N.Y.2d 232, 241, 677 N.Y.S.2d 764, 769, 700 N.E.2d 303, 308 (1998). I am not sure that reasonable minds could differ as to the extent of a store employee's knowledge of the hazard of placing his hand in the open spout of a meat-grinder, but I think it likely that the Court of Appeals would rule that reasonable minds might differ on the closely related question concerning the extent of the employee's knowledge of a safer alternative, i.e., avoiding use of a machine from which a safety guard had been removed and requesting a machine with the guard in place. For that reason, and in the absence of the clear answer we requested from the Court of Appeals, I agree that Liriano's case was properly submitted to a jury for its decision.

One final comment is warranted. Those who believe that every decision in human affairs is a rational one, influenced logically by the incentives and disincentives that inhere in a given set of circumstances, will think it perverse that a manufacturer can be liable for failure to warn about the hazard of a meat-grinder originally equipped with a safety guard that has subsequently been removed even though liability might not exist had no such guard been initially installed. Surely, the devout rationalists will say, a rule of law countenancing such seemingly contradictory results will create an incentive for meat-grinder manufacturers not to install safety guards in the first place, thereby obtaining at least the chance to escape liability that, under today's decision, is deemed appropriate for jury consideration. I acknowledge that the disincentive to install a safety guard might exist, but, as with many predictions made on the assumption that a disincentive to take action will result in the action not being taken (or that an incentive to take action will result in the action being taken), I think it is extremely doubtful that meat-grinder manufacturers will elect to forgo safety guards in the hope of avoiding failure-to-warn liability for meat-grinders from which such guards have been removed. We have been well advised that the life of the law is not logic but experience, see Oliver Wendell Holmes, Jr., The Common Law 1 (Little, Brown & Co. 1990)(1891), and it is often the case that the life of life itself is not logic. Though rationality guides many human actions, it does not guide them all. Despite the disincentive arguably created by the imposition of liability in this case, manufacturers might well elect to install safety guards simply because they have some concern (humanitarian, not economic) that hands should not be severed by their machines. Moreover, if our decision correctly predicts New York law, manufacturers of meat-grinders equipped with safety guards can readily avoid liability for injuries resulting from use after the guard has been removed by the inexpensive furnishing of some reasonable form of notice of the hazard of using the machine without the guard. Hobart has already acted in this direction by placing on its machine a warning against use if the safety guard has been removed. Thus, the circumstances giving rise to Hobart's liability in this case are unlikely to arise again.

For all of these reasons, I concur.