Ronnie Traylor and his wife appeal from the dismissal of their products liability suit, a diversity suit governed by Indiana law. The case was tried before a magistrate judge by consent of the parties and the jury brought in a verdict for the defendants, whom we refer to collectively as Omark.

A maul, the agent of the injury that gave rise to the suit, is a long-handled striking tool the steel head of which has an axe blade on one side and a flat, sledgehammer-like surface on the other. One day in 1986 Ronnie Traylor and his friend Dierking were splitting logs on Traylor's property. Each had a maul--Dierking a maul that had been manufactured by Omark. Each maul had been sold with a warning not to strike one maul against another, because chipping could occur and cause an eye injury, and not to use a maul without safety goggles. Traylor's maul got stuck in a log he was splitting. With the axe head embedded in the log, the flat surface of the head facing up, and the handle of the maul sticking out sideways, the two men set about to free the axe head. Traylor--who was not wearing safety glasses--crouched down, holding the handle of his maul to steady it, his unprotected face only a couple of feet from the head of the maul. Dierking gave the flat surface of Traylor's maul a whack with the flat surface of his maul. When Dierking's maul struck, it chipped, and the chip shot into Traylor's right eye and put the eye out.

Dierking had noticed before the accident that the flat surface of the head of his maul was chipped, cracked, and misshapen. There was considerable although not conclusive evidence that this was not due just to normal wear and tear--rather, that Dierking's maul was defective in two respects. First, it had an unusually narrow bevel around the flat surface of the head. A bevel is an angled as distinct from a right-angled edge. The larger the bevel, the less likely a surface is to chip. Second, the steel in the head, which had been manufactured in China, was of uneven hardness, which made it more likely that the head would chip when it was struck against another hard surface. The fact that a product is of low quality does not make it defective merely because higher-quality products are likely to be safer, safety being a dimension of quality. But Omark, while denying that its maul was defective, has not argued that the plaintiffs were trying to hold it to a standard of safety appropriate only to a more expensive maul.

Besides denying that the maul was defective, Omark raised affirmative defenses of misuse and "incurred risk," the latter a synonym for assumed risk. Dean v. Martz, 329 S.W.2d 371, 374 (Ky.1959); LaFata v. Busalaki, 291 S.W.2d 151, 154 (1956). The common law tort defense of "open and obvious danger" used to be available to defendants in products liability cases as in other personal injury cases in Indiana. Bemis Co. v. Rubush, 427 N.E.2d 1058, 1061 (Ind.1981); Estrada v. Schmutz Mfg. Co., 734 F.2d 1218 (7th Cir.1984). Then the Indiana legislature as part of a general overhaul of products liability law decided in effect to replace the defense of open and obvious danger (a much-criticized defense, for reasons sketched in id. at 1220), which the legislature abolished in products liability cases, with another common law defense, which the legislature proceeded to codify for those cases--that of assumed or incurred risk. Koske v. Townsend Engineering Co., 551 N.E.2d 437, 441-42 (Ind.1990); Montgomery Ward & Co. v. Gregg, 554 N.E.2d 1145, 1150-51 (Ind.App.1990). The products liability statute defines the defense as follows: "It is a defense that the user or consumer ... knew of the defect and was aware of the danger and nevertheless proceeded unreasonably to make use of the product and was injured by it." Ind.Code § 33-1-1.5-4(b)(1). Traylor was a bystander rather than a user or a consumer of the allegedly defective product--Dierking's maul--but the statute, including its defenses, applies to suits by foreseeable bystanders as well as to suits by users or consumers of the defective product. Ind.Code § 33-1-1.5-2; State Farm Fire & Casualty Co. v. Structo Division, 540 N.E.2d 597 (Ind.1989). The jury brought in a general verdict for the defendants, and as there was a good deal of evidence that the maul was indeed defective, and the applicability of the defense of misuse is obscure in a case like this (more on that later), it is entirely possible that the jury based its verdict on a determination that Traylor "knew of the defect" and decided to take his chances.

The doctrine of incurred risk, more familiarly assumption of risk, teaches that a person who proceeds in the face of what he knows to be a risk of a certain consequence cannot complain if the consequence materializes. Dierking may well have known that his maul was defective, so if he had been injured by it we may assume that he could not recover damages against Omark. But there is little or no evidence that Traylor knew that Dierking's maul was defective. He probably knew that striking a maul against another maul can cause one of the mauls to chip. The instructions that came with his own maul expressly warned him of this danger and he is (or at least was, before his accident) a factory worker who wore safety goggles at work to protect his eyes from metal chips. But to assume the risk of an eye injury from a flying maul chip is not the same thing as assuming the risk of an eye injury caused by a chip from a defective maul. These could be risks of different orders of magnitude. The difference is well illustrated by Moore v. Sitzmark Corp., 555 N.E.2d 1305 (Ind.App.1990), which involved safety binders for skis. A safety binder is designed to release the ski from the skier's boot when lateral force is applied to the binder, so that the boot (and the foot in it) won't be twisted by the ski. Anyone who skis knows that even the best safety binders are fallible, so that there is still a nontrivial risk of breaking one's leg. But of course if the safety binder is defective and therefore highly unlikely to release the ski in a fall, the danger of breaking one's leg is much greater. Moore holds that merely by going skiing and thus assuming some irreducible risk of being injured, the purchaser of safety binders does not assume the greater, hidden risk due to the defect in the particular binders he bought. If, however, he had known about the defect and had decided to take his chances, he would have assumed the specific risk of a defective binder and the doctrine of incurred risk would bar him from recovering damages if he was injured as a result of the defect.

It is the same here. If Traylor knew that Dierking's maul was defective and decided to take his chances, the incurred-risk statute bars recovery. But if he merely knew that striking two mauls together can cause one of them to chip even if neither is defective, that knowledge would not bar recovery. This is provided of course that the accident was due to the defect. If the accident would have occurred anyway, there would be no prima facie liability and the defense would not come into play.

All this is background to the question whether the nature of the incurred-risk defense was adequately explained to the jury. Instruction 20 explains that if the plaintiff "disregarded an adequate warning ... he incurred the risk of the injury he sustained," for "the doctrine of incurred risk is based upon the proposition that one incurs all the ordinary and usual risks of an act which he voluntarily enters, if those risks are known and understood by him." Instruction 21 explains that the "essential elements" of the defense are that Traylor knew either "of the danger associated with using a maul without adequate eye protection" or "of the danger associated with striking a maul with another maul," and that knowing these things he voluntarily exposed himself to the danger and was injured as a proximate result. These are the only instructions that deal with incurred risk. Nowhere do they state that the relevant knowledge is knowledge of the defect. The reference to "adequate warning" and to the two dangers associated with the use of a maul is a giveaway that the magistrate judge thought the defense applicable whenever the plaintiff knows that the use of the product in a particular way creates a danger of injury. That is not Indiana law, and would in fact strip plaintiffs of virtually all protection against hidden product defects.

Since there must be a new trial, it is not strictly necessary that we discuss the other errors allegedly committed by the magistrate judge, but we have decided to do so in order to provide guidance for the retrial and to minimize the probability of another appeal. One of these concerns instruction 18, which correctly instructed the jury on the two defects alleged by the Traylors but then went on to say that the jury was "not to consider the hardness of the maul." What the instruction doubtless meant, and probably what the jury understood it to mean in light of the testimony and the lawyer's arguments, was that the surface hardness (or softness) of the flat side of the head of Dierking's maul's was not alleged to be defective. The surface of the head had been hardened to acceptable industry specifications. The problem was that the interior of the head (the "microstructure," as it is called) was of uneven hardness, making the surface vulnerable (according to the plaintiffs' evidence) to chipping. We do not think the instruction, in context, was erroneous, but it would have been better if it had distinguished between surface and interior hardness, and this should be done in the next trial.

The next issue concerns the exclusion of evidence that Omark, after it had sold the maul in 1984 but before the accident occurred in 1986, had conducted tests on this model of maul which showed that the bevel was dangerously narrow, and also that Omark had specified in its purchase orders to the Chinese manufacturer a more uniform hardness of the microstructure. The ground of exclusion was Rule 407 of the Federal Rules of Evidence, which excludes evidence of remedial measures taken subsequent to "an event" if those measures are introduced into evidence for the purpose of showing "culpable conduct in connection with the event." We have understood "culpable conduct" to include the creation of a product defect. Flaminio v. Honda Motor Co., Ltd., 733 F.2d 463, 468-70 (7th Cir.1984); Probus v. K-Mart, Inc., 794 F.2d 1207 (7th Cir.1986). Omark argues that Rule 407 wasn't the ground of exclusion--Rule 402 (relevance) was. It is true that the magistrate judge did not mention Rule 407. But Rule 407 was the ground on which Omark sought exclusion, and whatever else the evidence was or was not, it was relevant to the question whether the maul was defective.

The problem with applying Rule 407 was not a lack of culpable conduct but the fact that the remedial measures were taken before rather than after the "event," which in an accident case the courts have invariably and we think correctly understood to mean the accident. Dukett v. Mausness, 546 N.E.2d 1292 (Ind.App.1989); Kaczmarek v. Allied Chemical Corp., 836 F.2d 1055, 1060 (7th Cir.1987); Oberst v. International Harvester Co., 640 F.2d 863, 866 (7th Cir.1980). The magistrate judge understood "event" to mean the time at which the allegedly defective product passed out of the hands of the defendant, that is, when Omark sold it. This is not an absurd interpretation, although we cannot understand its application to the tests, as distinct from the specifying of a safer design. Passing that point, we agree that the object of the rule, which is to encourage safety improvements by preventing their being used as evidence to show that the defendant should have made the improvements earlier, is undermined whenever improvements made subsequent to the creation of whatever dangerous condition gave rise to an accident are used as evidence that the condition was culpable. But in deciding how far back to push "event" one must keep in mind the desirability not only of encouraging safety improvements but also of permitting the finder of fact, here a jury, to consider probative evidence, as the evidence in question undoubtedly was. The latter consideration weighs more heavily when the improvements are made before an accident occurs. For then it is open to the producer to limit his liability by recalling the unimproved product. Ordinarily, indeed, this will be the prudent course--to correct a defect before it causes accidents for which the manufacturer will be liable. Which makes us doubt that a producer will often be deflected from making improvements by fear about the consequences to his litigating position in hypothetical future cases--cases especially hypothetical because no accident has yet occurred. We add that even if the magistrate judge had been correct to exclude the evidence in question from the plaintiffs' case in chief, he had no basis for excluding its use to impeach testimony by Omark's expert witnesses. Rule 407 expressly permits the use of evidence of subsequent repairs for purposes of impeachment.

Omark had run tests on the maul heads which showed that a high percentage did not come up to the company's own standards of uniform hardness. The magistrate judge excluded these tests, as being relevant only to punitive damages; for he had bifurcated the trial into a liability and a damages phase. Apparently the plaintiffs had intended to use the evidence primarily to show that Omark knew the maul was defective, which is separate from whether it was defective; but we think they made adequately clear that they wanted to use the evidence to show the existence of a defect as well. Omark argues that the evidence cannot show this because there is no indication that Dierking's maul was one of those that was tested and found wanting. The objection is shallow. The point is that the Chinese maul heads did not come up to Omark's own standards for minimally uniform hardness, and this made it somewhat more likely that they did not come up to a standard of reasonably uniform hardness--the legal standard--either.

We can pass by the plaintiffs' objection to bifurcating the trial after the jury was sworn, a procedure the plaintiffs argue was too abrupt and interfered with their ability to marshal their evidence because they had to disentangle their liability evidence from their damages evidence. That problem is unlikely to recur at the new trial; the plaintiffs are on notice now that the trial may be bifurcated, although the decision on bifurcation will presumably rest with the judge (a different one, 7th Cir.R. 36) who conducts the second trial. There is a possibility, though not a large one, that another problem in the first trial will recur in the second. After one of the defendant's expert witnesses gave his testimony in the usual way, which was on a Friday, he announced that he couldn't appear in court on Monday, when the trial was to resume, to be cross-examined, because he had a prior engagement. The magistrate judge decided that the witness would be cross-examined on Saturday. The jury would not be there but the cross-examination would be videotaped and the videotape shown to the jury on Monday. This was done. The plaintiffs argue that instead the witness should have been recalled on Tuesday or Wednesday so that he could have been cross-examined in person. Although we have no objection to videotaped testimony and do not believe that the fact that this witness's direct testimony was live and his cross-examination taped was a reversible error, we do think this sort of "dual media" testimony is generally, and was in this instance, a bad idea. Psychologists and decision theorists point out, what is anyway common sense, that a living person generally conveys a stronger impression than does his resume, or a transcript of his remarks. Richard E. Nisbett et al., "Popular Induction: Information Is Not Necessarily Informative," in Judgment Under Uncertainty: Heuristics and Biases 101, 111-15 (Daniel Kahneman, Paul Slovic & Amos Tversky ed. 1982). On this ground they warn for example against the tendency of employers to give impressions gleaned from interviews with job applicants much greater weight than impressions gleaned from written applications. By presenting its expert witness's direct testimony live but his cross-examination taped, Omark was able to give artificially greater salience to the part of his examination that favored Omark than to the part that favored its opponent. There was a thumb on the scale. It should be removed in the retrial.

The last issue is whether the plaintiffs were entitled to a directed verdict on the defense of misuse raised by Omark. The plaintiffs argue with some basis in the statute that the only relevant misuse is that which is unforeseeable by the producer, and there is plenty of evidence that the dangerous practice of striking two mauls together is common and that this was well known to Omark--in fact spurred it to warn purchasers of its mauls explicitly against the practice. The products liability statute does not define misuse. Ind.Code § 33-1-1.5-4(b)(2). But it has a well-recognized meaning in products liability law. It means to use a product in a way in which it was not intended to be used. Montgomery Ward & Co. v. Gregg, supra, 554 N.E.2d at 1151-52 and n. 2, 1156; Estrada v. Schmutz Mfg. Co., supra, 734 F.2d at 1220-21; W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 102 at pp. 711-12 (5th ed. 1984). If the buyer or user of a product uses it in a way in which it was not intended to be used, and he knows this, and as a result of his unintended and unauthorized use he is injured, he has--prima facie--only himself to blame.

We said "prima facie." If the producer knows that his product is being widely misused, he may have a duty to make reasonable efforts to warn against the consequences of the misuse. This suggestion, which is consistent with the statute's confinement of the defense to cases in which the misuse is "not reasonably expected by the seller at the time of sale," and is supported by Indiana case law, Conder v. Hull Lift Truck, Inc., 435 N.E.2d 10, 17 (Ind.1982); id. at 20 (dissenting opinion); Craven v. Niagara Machine & Tool Works, Inc., 417 N.E.2d 1165, 1170 (Ind.App.1981), is a refinement of the balancing of responsibilities for care that is at the heart of tort law. On the one hand, the concept of misuse implies that the (mis)user could have avoided the accident just by confining his use of the product to the intended use. But Americans are not Prussians--they are not schooled to obedience as the prime virtue of the good citizen--so they are constantly putting things to new uses. If they have no reason to think the new use unreasonably dangerous, and therefore reasonably believe that the benefit of the deviant use exceeds its cost, their culpability in disobeying the instructions is slight. If the producer can protect the user at lower cost by a simple warning, this may be the cheapest method for the prevention of accidents. Although a maul is intended for splitting logs rather than for pounding other mauls, it is the most natural thing in the world, if you find yourself with one maul stuck in a log, to whack it with another maul in an effort to make the first complete the splitting of the log in which it is stuck, and thus get free. If there are hidden dangers in such a procedure, optimal accident avoidance may require the producer of the maul to warn of these dangers. Id. As Omark did.

What all this has to do with the present case eludes our understanding, however. Insofar as there is an inherent danger of chipping when one maul hits another, not only did Omark warn of the danger but the plaintiffs do not argue that this inherent danger would have made the maul defective in the absence of a warning. The defects they allege are different. And we have seen that there was no warning against them. So we are puzzled why misuse became an issue in this case, but this is something that can be explored further on remand should a new trial actually be necessary to resolve this lawsuit. Maybe Omark is trying to argue that a statement of unauthorized use insulates it from liability for any risks or consequences of that use. Whether that is its argument, and whether, if so, it has any merit, we need not decide on this appeal.

The judgment is reversed with directions to grant the plaintiffs a new trial to be conducted in conformity with this opinion.

REVERSED AND REMANDED, WITH DIRECTIONS.