POSNER, Chief Judge.

Maria Navarro, the plaintiff in a diversity personal-injury suit which the parties agree is governed by the law of Illinois, was a passenger in a 1982 Subaru that had been manufactured in Japan in 1981 by Fuji Heavy Industries, the defendant. The driver lost control of the car when its rear suspension gave way suddenly and unexpectedly as a result of having rusted through. Navarro was flung from the car, which rolled over several times and came to rest on top of her. The accident occurred in 1992 in Chicago, where the car, which had 125,000 miles on its odometer, had mainly been driven. Her claim of strict products liability was barred by Illinois' 10-year statute of repose for such suits, 735 ILCS 5/13-213(b), so Navarro proceeded on the basis of a claim of negligence that Fuji concedes is not barred by the statute of repose. The claim is that Fuji was negligent in designing the car's rear suspension in such a way that it might rust through if the car was driven in an area in which the roads are heavily salted during the winter — as is the case in Chicago — and, moreover, rust through from the inside out (the suspension not being completely sealed against the elements) so that there would be no visual clue that the suspension was about to give way. Two years before the accident, Fuji had acknowledged the problem in a notice of recall that the owner-driver of the Subaru in which Navarro was injured had not seen. Fuji does not argue that the sending of the recall notice to known owners of its cars insulates it from liability, and, conversely, Navarro does not argue that Fuji breached a duty to warn owners of the defect once Fuji discovered it. After the case was removed from the Illinois state court in which it had been filed originally, the district judge granted summary judgment for Fuji on the ground that Navarro had failed to present sufficient evidence of defective design to create a triable issue, and he dismissed the suit. 925 F. Supp. 1323 (N.D. Ill. 1996). Later he dismissed the plaintiff's motion for reconsideration as well.

As we said, this suit is based on negligence rather than on strict products liability. But there is little or no practical difference in a case of defective design, at least so far as the standard of liability is concerned (we have just seen that there is a big difference with respect to the deadline for bringing suit): you must prove that the design was defective in either kind of case, and whether the design was defective is determined by use of the same Hand-formula or cost-benefit approach that is used to determine negligence in a tort case not involving a product. E.g., Rosen v. Ciba-Geigy Corp., 78 F.3d 316, 317 (7th Cir. 1996); Bammerlin v. Navistar International Transportation Corp., 30 F.3d 898, 902 (7th Cir. 1994); Flaminio v. Honda Motor Co., 733 F.2d 463, 467 (7th Cir. 1984). It is true that cases in Illinois and elsewhere often say that in a strict liability products case the manufacturer's negligence is irrelevant. E.g., Simpson v. General Motors Corp., 455 N.E.2d 137, 141 (Ill.App. 1983), aff'd, 483 N.E.2d 1 (Ill. 1985). But the significance of such language is in curtailing the defense of contributory negligence in products cases and in preventing the manufacturer from defending such a case on the ground that the negligence in the design was due to carelessness by the manufacturer of a component. Id. at 141; Flaminio v. Honda Motor Co., supra, 733 F.2d at 467. The points are related: the manufacturer cannot defend such a case by pointing to someone else's negligence, whether the victim's or a supplier's. It is also true that the cases frequently offer the plaintiff a choice between proving that the design was defective and proving that it was not as safe as the consumer would reasonably have expected. E.g., Lamkin v. Towner, 563 N.E.2d 449, 457 (Ill. 1990). But this comes to the same thing; the consumer expects the products he buys not to be defectively designed.

So there is one test, and it is objective, Flaminio v. Honda Motor Co., supra, 733 F.2d at 467; it is based on what Fuji should have done about rust in light of what it should have known, not what it should have done in light of what it did know. The fact that as a Japanese firm it may not have known much about Chicago driving conditions is irrelevant (as well as highly improbable, since it has and had an elaborate distribution and dealership network throughout the United States, including the Chicago area). Fuji intended to sell its cars in the Chicago area and is therefore charged with whatever knowledge can reasonably be imputed to automobile manufacturers, wherever located, desiring to serve that market. W. Page Keeton et al., Prosser and Keeton on the Law of Torts sec. 32, pp. 184 (5th ed. 1984).

And while we are clearing away the underbrush of this suit we point out that there is no merit to Fuji's argument, which it puts forward as an alternative ground for upholding the district court's judgment, that the Subaru "just wore out," so that the accident was not caused by the defect in the suspension. It is true that cars become inoperable for a variety of reasons many of which are not attributable to the manufacturer, including driver abuse and neglect, poor maintenance by dealers and auto repair shops, defective replacement parts, collisions, metal fatigue, and corrosion. Fuji did not warrant that its 1982 Subarus would still be operable after ten years and 125,000 miles of driving on Chicago's roads — but neither did it warn buyers that it was unsafe to drive a car with such a history because it might veer out of the driver's control without warning. Nor did the car fall apart because of the mechanical equivalent of old age (the one-horse-shay phenomenon) or the actions or inactions of the driver, the dealer, or anyone else — besides the designers of the rear suspension system. Had they designed it differently, the car would not have given out as soon as it did (at least without warning from visible rust) and Navarro would not have been injured. Fuji does not argue that the car was otherwise in so dangerous a condition through no fault of Fuji's that Navarro's damages should be cut down to reflect the likelihood that she would soon have been injured by some other disastrous failure of the car's aging components. Lancaster v. Norfolk Western Ry., 773 F.2d 807, 822-23 (7th Cir. 1985); Stoleson v. United States, 708 F.2d 1217, 1223-24 (7th Cir. 1983); Steinhauser v. Hertz Corp., 421 F.2d 1169, 1173-74 (2d Cir. 1970) (Friendly, J.); cf. Doll v. Brown, 75 F.3d 1200, 1205-06 (7th Cir. 1996). On this record we must assume that the design of the rear suspension was a cause of the accident, and a culpable cause if the design was defective.

To show that it was defective — that is, that due care required Fuji to design the rear suspension in such a way that it either would not rust even if driven for many years in Chicago or, more plausibly, that if it did rust the rust would show so that the suspension could be replaced before it gave way — the plaintiff presented the affidavit of a qualified engineer, the recall notice, a corrosion report by a trade association, and an affidavit by another expert that the judge properly ruled inadmissible because unverified and that in any event added nothing material to the first expert's affidavit. The affidavit of the first expert was the centerpiece of Navarro's case and the judge found that it was inadmissible under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993). See also People Who Care v. Rockford Board of Education, 111 F.3d 528, 537 (7th Cir. 1997); Minasian v. Standard Chartered Bank, PLC, 109 F.3d 1212, 1216 (7th Cir. 1997); Cummins v. Lyle Industries, 93 F.3d 362, 367-69 (7th Cir. 1996); Rosen v. Ciba-Geigy Corp., supra, 78 F.3d at 318-19.

It is noteworthy that the affidavit and the plaintiff's other evidentiary materials were not presented in response to affidavits or other evidence in support of the defendant's motion for summary judgment. The defendant did not present any evidence in support of its motion. It was not required to. Celotex Corp. v. Catrett, 477 U.S. 317, 324 (1986); Fitzpatrick v. Catholic Bishop, 916 F.2d 1254, 1256 (7th Cir. 1990). It could put the plaintiff to her proof, though in doing so it was taking the risk that the plaintiff might try to establish her prima facie case not by presenting evidence but by invoking the doctrine of res ipsa loquitur, which in Illinois goes by the name of the Tweedy doctrine when it is invoked in a products liability case. Tweedy v. Wright Ford Sales, Inc., 357 N.E.2d 449, 452 (Ill. 1976); see Rizzo v. Corning Inc., 105 F.3d 338, 340 (7th Cir. 1997); Welge v. Planters Lifesavers Co., 17 F.3d 209, 211 (7th Cir. 1994). The plaintiff might have argued that you don't need expert evidence to show that it is negligent to design a rear suspension system in such a way that if it rusts, the rust will be invisible until the system suddenly collapses. We don't say that such an argument would fly (probably it would not, as will become clearer), but only that it wasn't attempted; the plaintiff has tacitly conceded that in order to withstand summary judgment she had to present evidence that Fuji was negligent in designing the suspension system as it did.

The district judge was only half right in finding that the expert's affidavit flunked the Daubert test. The expert is an experienced consultant in the relevant fields of failure analysis, mechanical safety, and accident reconstruction; he inspected the ruined car; and he explained lucidly how the accident had come about — through the insidious corrosion of the rear suspension. So far, so good; there was no basis for excluding this part of the affidavit. Wheeler v. John Deere Co., 935 F.2d 1090, 1100-01 (10th Cir. 1991). The problem is that the expert didn't stop there. First, he observed correctly but inconsequently that Fuji was aware of the special corrosion problem in Chicago. So far as appears the expert was here merely paraphrasing the recall notice. The notice was issued in 1990, and without more does not show that Fuji or any other automobile manufacturer knew or should have known about the problem in 1981, when the car was manufactured, or 1982, when it was sold. (The same problem afflicts evidence Navarro tried to present of identical accidents that occurred after 1982.) No doubt American manufacturers, at least — and remember that Fuji cannot plead excusable ignorance on the basis of its being a foreign company — knew that road conditions in Chicago are more corrosive than elsewhere, but it cannot be assumed that they knew they should be worrying about the corrosion's starting inside the suspension.

A more fundamental point is that knowledge of a hazard does not equate to negligence for failing to take a particular precaution against it. Negligence depends on the magnitude and likelihood of the hazard and on the cost of preventing the hazard or reducing its likelihood. E.g., Yager v. Illinois Bell Tel. Co., 667 N.E.2d 1088, 1092 (Ill.App. 1996); Hendricks v. Peabody Coal Co., 253 N.E.2d 56, 61 (Ill.App. 1969); Bammerlin v. Navistar International Transportation Corp., supra, 30 F.3d at 902; McCarty v. Pheasant Run, Inc., 826 F.2d 1554, 1557 (7th Cir. 1987). Remember too that the specific hazard at issue in this case is not the rusting of the suspension per se but its rusting in a manner that would supply no visual clue until too late; and there is no indication in the affidavit that Fuji or anyone else knew or should have known of this hazard in 1981.

The affidavit goes on to state that "prior to 1981, Fuji Heavy Industries knew of the corrosive effects of road salt on the suspension arms of a vehicle" — but this amounts to nothing more than the truism that a manufacturer of a steel product is charged with elementary knowledge that salt is corrosive. With little more ado the expert concluded that "the design and manufacture of the rear suspension arm in 1981 was negligent and unreasonable." The affidavit contains no support for this conclusion, and a conclusion without any support is not one based on expert knowledge and entitled to the dignity of evidence. Daubert v. Merrell Dow Pharmaceuticals, Inc., supra, 509 U.S at 590; Deimer v. Cincinnati Sub-Zero Products, Inc., 58 F.3d 341, 344-45 (7th Cir. 1995); Bammerlin v. Navistar International Transportation Corp., supra, 30 F.3d at 901. Thus at most the affidavit is evidence that Fuji should have known back in 1981 that its rear suspension might rust through in highly corrosive driving environments. There is no indication that responsible engineers at the time thought the hazard considerable because undetectable, and therefore worth guarding against and if so by what means, or that Fuji could have redesigned the suspension to avoid the hazard at a cost commensurate with the expected accident cost, which might reasonably have seemed slight — or indeed at any cost. The affidavit is silent on what other automobile manufacturers were doing about corrosion in 1981. All that one can draw out of the affidavit is that the accident was indeed caused by the undetectable rusting of the rear suspension and that Fuji knew that the suspension might rust — but not that it might do so undetectably — if driven on heavily salted roads.

The plaintiff argues that the gaps in the expert's affidavit would have been filled if only the defendant, before moving for summary judgment, had deposed the expert. But there is no duty to cross-examine or depose your opponent's witnesses so that they can supplement the testimony they failed to give on direct examination or in their affidavit. An expert's affidavit must be sufficiently complete to satisfy the criteria of the Daubert decision, and one of those criteria, as we have been at pains to emphasize, is that the expert show how his conclusion, in this case that Fuji should in the exercise of due care have designed the 1982 Subaru's rear suspension differently, is grounded in — follows from — an expert study of the problem.

The recall notice adds nothing to the affidavit. It merely shows that in 1990 (for there is no indication that Fuji discovered the problem earlier and sat on it) Fuji discovered the problem of the undetectable rusting. The only other admissible document on which the plaintiff relied in opposing summary judgment is an undated report of the American Iron and Steel Institute on corrosion. (Internal evidence suggests that the report was written in the early years of this decade.) The report points out that because of the high cost of electricity in Japan, Japanese auto manufacturers use lighter alloys to rustproof their cars than American manufacturers do but that in view of the corrosive conditions on many American roads the Japanese manufacturers are moving toward heavier alloys. This is some evidence that in the early 1980s Japanese cars were not as resistant to rust as American cars were, but it is not, without more — and there is no more — evidence that the difference created an incremental hazard great enough to make the Japanese cars defective.

When the inadmissible, because nakedly conclusional, portions of the expert's affidavit are stricken, there just is no evidence, though there may be a common-sense suspicion, that the 1982 Subaru's rear suspension was defective. The district judge was therefore right to grant summary judgment for Fuji. The plaintiff, however, moved for reconsideration under Fed.R.Civ.P. 59(e) and attached a new, amended affidavit of its expert, plus some answers to interrogatories summarizing additional expert testimony, interrogatories that the plaintiff had made a part of the record before the suit had been removed to federal court but had failed to bring to the district judge's attention — wisely so. For the most arresting answer is an acknowledgment that not until 1985, four years after the manufacture of the fatal vehicle, had Chicago been classified as having severely corrosive road conditions. That was in the AISI report that we mentioned; there is no indication that anyone else does these classifications.

The amended report comes a little closer to adequacy, mainly by citing engineering literature from the 1970s that, according to the affidavit, shows that the corrosion problem in Chicago was well known then and that there were accepted methods of avoiding the hazard that caused the 1982 Subaru to fall apart when it did. The literature is not appended to the affidavit or explained in it; it is merely cited; but we won't quibble, and will assume for the sake of argument that had it been placed in evidence before the judge ruled on the defendant's motion for summary judgment he would have had to deny the motion. The insuperable difficulty for the plaintiff is that a motion to alter the judgment may not be based on evidence that was available when the district judge took the motion for summary judgment under advisement but that was not presented then. Retired Chicago Police Ass'n v. City of Chicago, 76 F.3d 856, 867 (7th Cir. 1996); In re Prince, 85 F.3d 314, 324 (7th Cir. 1996). If the plaintiff needed more time to obtain a decent affidavit, she could have asked the district judge for more time, Fed.R.Civ.P. 56(f); Celotex Corp. v. Catrett, supra, 477 U.S. at 326, and she did not. He had no reason to believe that she was holding evidence back, and no obligation to give her two bites at the apple. If district judges were required to consider evidence newly presented but not newly discovered after judgment, there would be two rounds of evidence in a great many cases. Even in this court, the plaintiff has offered no reason for her failure to submit a better affidavit on the first round — except that she expected the defendant to give the expert a chance to plug the holes in his testimony by deposing him.

AFFIRMED.