Richard Welge, forty-something but young in spirit, loves to sprinkle peanuts on his ice cream sundaes. On January 18, 1991, Karen Godfrey, with whom Welge boards, bought a 24-ounce vacuum-sealed plastic-capped jar of Planters peanuts for him at a K-Mart store in Chicago. To obtain a $2 rebate that the maker of Alka-Seltzer was offering to anyone who bought a "party" item, such as peanuts, Godfrey needed proof of her purchase of the jar of peanuts; so, using an Exacto knife (basically a razor blade with a handle), she removed the part of the label that contained the bar code. She then placed the jar on top of the refrigerator, where Welge could get at it without rooting about in her cupboards. About a week later, Welge removed the plastic seal from the jar, uncapped it, took some peanuts, replaced the cap, and returned the jar to the top of the refrigerator, all without incident. A week after that, on February 3, the accident occurred. Welge took down the jar, removed the plastic cap, spilled some peanuts into his left hand to put on his sundae, and replaced the cap with his right hand--but as he pushed the cap down on the open jar the jar shattered. His hand, continuing in its downward motion, was severely cut, and is now, he claims, permanently impaired.

Welge brought this products liability suit in federal district court under the diversity jurisdiction; Illinois law governs the substantive issues. Welge named three defendants (plus the corporate parent of one--why we don't know). They are K-Mart, which sold the jar of peanuts to Karen Godfrey; Planters, which manufactured the product--that is to say, filled the glass jar with peanuts and sealed and capped it; and Brockway, which manufactured the glass jar itself and sold it to Planters. After pretrial discovery was complete the defendants moved for summary judgment. The district judge granted the motion on the ground that the plaintiff had failed to exclude possible causes of the accident other than a defect introduced during the manufacturing process.

No doubt there are men strong enough to shatter a thick glass jar with one blow. But Welge's testimony stands uncontradicted that he used no more than the normal force that one exerts in snapping a plastic lid onto a jar. So the jar must have been defective. No expert testimony and no fancy doctrine are required for such a conclusion. A nondefective jar does not shatter when normal force is used to clamp its plastic lid on. The question is when the defect was introduced. It could have been at any time from the manufacture of the glass jar by Brockway (for no one suggests that the defect might have been caused by something in the raw materials out of which the jar was made) to moments before the accident. But testimony by Welge and Karen Godfrey, if believed--and at this stage in the proceedings we are required to believe it--excludes all reasonable possibility that the defect was introduced into the jar after Godfrey plucked it from a shelf in the K-Mart store. From the shelf she put it in her shopping cart. The checker at the check-out counter scanned the bar code without banging the jar. She then placed the jar in a plastic bag. Godfrey carried the bag to her car and put it on the floor. She drove directly home, without incident. After the bar-code portion of the label was removed, the jar sat on top of the refrigerator except for the two times Welge removed it to take peanuts out of it. Throughout this process it was not, so far as anyone knows, jostled, dropped, bumped, or otherwise subjected to stress beyond what is to be expected in the ordinary use of the product. Chicago is not Los Angeles; there were no earthquakes. Chicago is not Amityville either; no supernatural interventions are alleged. So the defect must have been introduced earlier, when the jar was in the hands of the defendants.

But, they argue, this overlooks two things. One is that Karen Godfrey took a knife to the jar. And no doubt one can weaken a glass jar with a knife. But nothing is more common or, we should have thought, more harmless than to use a knife or a razor blade to remove a label from a jar or bottle. People do this all the time with the price labels on bottles of wine. Even though mishandling or misuse, by the consumer or by anyone else (other than the defendant itself), is a defense, though a limited and (subject to a qualification noted later) partial defense, to a products liability suit in Illinois as elsewhere, e.g., J.I. Case Co. v. McCartin-McAuliffe Plumbing & Heating, Inc., 118 Ill. 2d 447, 114 Ill. Dec. 105, 111, 516 N.E.2d 260, 266 (1987); King v. American Food Equipment Co., 160 Ill.App.3d 898, 112 Ill. Dec. 349, 356, 513 N.E.2d 958, 965 (1987); Early-Gary, Inc. v. Walters, 294 So. 2d 181, 186-87 (Miss.1974); Annot., "Products Liability: Sufficiency of Evidence to Support Product Misuse Defense in Actions Concerning Bottles, Cans, Storage Tanks, or Other Containers," 58 A.L.R. 4th 160 (1987), and even if, as we greatly doubt, such normal mutilation as occurred in this case could be thought a species of mishandling or misuse, a defendant cannot defend against a products liability suit on the basis of a misuse that he invited. The Alka-Seltzer promotion to which Karen Godfrey was responding when she removed a portion of the label of the jar of Planters peanuts was in the K-Mart store. It was there, obviously, with K-Mart's permission. By the promotion K-Mart invited its peanut customers to remove a part of the label on each peanut jar bought, in order to be able to furnish the maker of Alka-Seltzer with proof of purchase. If one just wants to efface a label one can usually do that by scraping it off with a fingernail, but to remove the label intact requires the use of a knife or a razor blade. Invited misuse is no defense to a products liability claim. Invited misuse is not misuse.

The invitation, it is true, was issued by K-Mart, not by the other defendants; and we do not know their involvement, if any, in the promotion. As to them, the defense of misuse must fail, at this stage of the proceedings, for two other reasons. The evidence does not establish with the certitude required for summary judgment that the use of an Exacto knife to remove a label from a jar is a misuse of the jar. And in a regime of comparative negligence misuse is not a defense to liability but merely reduces the plaintiff's damages, unless the misuse is the sole cause of the accident.

Even so, the defendants point out, it is always possible that the jar was damaged while it was sitting unattended on the top of the refrigerator, in which event they are not responsible. Only if it had been securely under lock and key when not being used could the plaintiff and Karen Godfrey be certain that nothing happened to damage it after she brought it home. That is true--there are no metaphysical certainties--but it leads nowhere. Elves may have played ninepins with the jar of peanuts while Welge and Godfrey were sleeping; but elves could remove a jar of peanuts from a locked cupboard. The plaintiff in a products liability suit is not required to exclude every possibility, however fantastic or remote, that the defect which led to the accident was caused by someone other than one of the defendants. The doctrine of res ipsa loquitur teaches that an accident that is unlikely to occur unless the defendant was negligent is itself circumstantial evidence that the defendant was negligent. The doctrine is not strictly applicable to a products liability case because unlike an ordinary accident case the defendant in a products case has parted with possession and control of the harmful object before the accident occurs. St. Paul Fire & Marine Ins. Co. v. Michelin Tire Corp., 12 Ill.App.3d 165, 298 N.E.2d 289, 297-98 (1973). But the doctrine merely instantiates the broader principle, which is as applicable to a products case as to any other tort case, that an accident can itself be evidence of liability. Id., 298 N.E.2d at 298; Doyle v. White Metal Rolling & Stamping Corp., 249 Ill.App.3d 370, 188 Ill. Dec. 339, 346 and n. 3, 618 N.E.2d 909, 916 and n. 3 (1993). If it is the kind of accident that would not have occurred but for a defect in the product, and if it is reasonably plain that the defect was not introduced after the product was sold, the accident is evidence that the product was defective when sold. The second condition (as well as the first) has been established here, at least to a probability sufficient to defeat a motion for summary judgment. Normal people do not lock up their jars and cans lest something happen to damage these containers while no one is looking. The probability of such damage is too remote. It is not only too remote to make a rational person take measures to prevent it; it is too remote to defeat a products liability suit should a container prove dangerously defective.

Of course, unlikely as it may seem that the defect was introduced into the jar after Karen Godfrey bought it if the plaintiffs' testimony is believed, other evidence might make their testimony unworthy of belief--might even show, contrary to all the probabilities, that the knife or some mysterious night visitor caused the defect after all. The fragments of glass into which the jar shattered were preserved and were examined by experts for both sides. The experts agreed that the jar must have contained a defect but they could not find the fracture that had precipitated the shattering of the jar and they could not figure out when the defect that caused the fracture that caused the collapse of the jar had come into being. The defendants' experts could neither rule out, nor rule in, the possibility that the defect had been introduced at some stage of the manufacturing process. The plaintiff's expert noticed what he thought was a preexisting crack in one of the fragments, and he speculated that a similar crack might have caused the fracture that shattered the jar. This, the district judge ruled, was not enough.

But if the probability that the defect which caused the accident arose after Karen Godfrey bought the jar of Planters peanuts is very small--and on the present state of the record we are required to assume that it is--then the probability that the defect was introduced by one of the defendants is very high. In principle there is a third possibility--mishandling by a carrier hired to transport the jar from Brockway to Planters or Planters to K-Mart--but we do not even know whether a carrier was used for any of these shipments, rather than the shipper's own trucks. Apart from that possibility, which has not been mentioned in the litigation so far and which in any event, as we are about to see, would not affect K-Mart's liability, the jar was in the control of one of the defendants at all times until Karen Godfrey bought it.

Which one? It does not matter. The strict-liability element in modern products liability law comes precisely from the fact that a seller subject to that law is liable for defects in his product even if those defects were introduced, without the slightest fault of his own for failing to discover them, at some anterior stage of production. Crowe v. Public Building Comm'n, 74 Ill. 2d 10, 23 Ill. Dec. 80, 81, 383 N.E.2d 951, 952 (1978); Thomas v. Kaiser Agricultural Chemicals, 81 Ill. 2d 206, 40 Ill. Dec. 801, 805, 407 N.E.2d 32, 36 (1980); Flaminio v. Honda Motor Co., 733 F.2d 463, 467 (7th Cir.1984). So the fact that K-Mart sold a defective jar of peanuts to Karen Godfrey would be conclusive of K-Mart's liability, and since it is a large and solvent firm there would be no need for the plaintiff to look further for a tortfeasor. This point seems to have been more or less conceded by the defendants in the district court--the thrust of their defense was that the plaintiff had failed to show that the defect had been caused by any of them--though this leaves us mystified as to why the plaintiff bothered to name additional defendants.

And even if, as we doubt, the plaintiff took on the unnecessary burden of proving that it is more likely than not that a given defendant introduced the defect into the jar, he might be able to avail himself of the rule of Ybarra v. Spangard, 25 Cal. 2d 486, 154 P.2d 687 (1944), and force each defendant to produce some exculpatory evidence. Hessel v. O'Hearn, 977 F.2d 299, 305 (7th Cir.1992). In fact K-Mart put in some evidence on the precautions it takes to protect containers of food from being damaged by jarring or bumping. A jury convinced by such evidence, impressed by the sturdiness of jars of peanuts (familiar to every consumer), and perhaps perplexed at how the process of filling a jar with peanuts and vacuum-sealing it could render a normal jar vulnerable to collapsing at a touch, might decide that the probability that the defect had been introduced by either K-Mart or Planters was remote. So what? Evidence of K-Mart's care in handling peanut jars would be relevant only to whether the defect was introduced after sale; if it was introduced at any time before sale--if the jar was defective when K-Mart sold it--the source of the defect would be irrelevant to K-Mart's liability. In exactly the same way, Planters' liability would be unaffected by the fact, if it is a fact, that the defect was due to Brockway rather than to itself. To repeat an earlier and fundamental point, a seller who is subject to strict products liability is responsible for the consequences of selling a defective product even if the defect was introduced without any fault on his part by his supplier or by his supplier's supplier.

In reaching the result she did the district judge relied heavily on Erzrumly v. Dominick's Finer Foods, Inc., 50 Ill.App.3d 359, 8 Ill. Dec. 446, 365 N.E.2d 684 (1977). A six-year-old was injured by a Coke bottle that she was carrying up a flight of stairs to her family's apartment shortly after its purchase. The court held that the plaintiff had failed to eliminate the possibility that the Coke bottle had failed because of something that had happened after it left the store. If, as the defendants in our case represent, the bottle in Erzrumly "exploded," that case would be very close to this one. A nondefective Coke bottle is unlikely to explode without very rough handling. The contents are under pressure, it is true, but the glass is strengthened accordingly. But it was unclear in Erzrumly what had happened to the bottle. There was testimony that the accident had been preceded by the sound of a bottle exploding but there was other evidence that the bottle may simply have been dropped and have broken--the latter being the sort of accident that happens commonly after purchase. Although the opinion contains some broad language helpful to the defendants in the present case, the holding was simply that murky facts required the plaintiff to make a greater effort to determine whether the product was defective when it left the store. Here we know to a virtual certainty (always assuming that the plaintiff's evidence is believed, which is a matter for the jury) that the accident was not due to mishandling after purchase but to a defect that had been introduced earlier.

Even the narrow holding of Erzrumly is probably wrong, in light of bottle and other container cases decided by Illinois courts both before and after, Tweedy v. Wright Ford Sales, Inc., 64 Ill. 2d 570, 2 Ill. Dec. 282, 357 N.E.2d 449 (1976); Mabee v. Sutliff & Case Co., 404 Ill. 27, 88 N.E.2d 12 (Ill.1949); Fullreide v. Midstates Beverage Co., 70 Ill.App.3d 758, 27 Ill. Dec. 107, 388 N.E.2d 1070 (1979); Roper v. Dad's Root Beer Co., 336 Ill.App. 91, 82 N.E.2d 815 (1948), as well as by courts of other states. E.g., Van Duzer v. Shoshone Coca Cola Bottling Co., 103 Nev. 383, 741 P.2d 811 (1987) (per curiam); Virgil v. "Kash N' Karry" Service Corp., 61 Md.App. 23, 484 A.2d 652, 657 (Md.App.1984); Renninger v. Foremost Dairies, Inc., 171 So. 2d 602, 604 (Fla.App.1965). Right or wrong, Erzrumly is plainly contrary to Fullreide; and obviously when state courts of the same level reach opposite conclusions, a federal court in a diversity case is not bound to follow either.

REVERSED AND REMANDED.