This is a product liability action brought under the theory of implied warranty, which is a form of strict liability in tort. The plaintiff’s burden of proof for breach of implied warranty, as stated by this court “* * * consists of alleging and proving, by a preponderance of the evidence, that: (1) There was, in fact, a defect in the product manufactured and sold by the defendant; (2) such defect existed at the time the product left the hands of the defendant; and (3) the defect was the direct and proximate cause of the plaintiff’s injuries or loss.” State Auto Mutual Ins. Co. v. Chrysler Corp. (1973), 36 Ohio St. 2d 151. In the instant case, there is no question that, if a defect in the mounting of the steering column has been proved, there is sufficient indication of probable cause to allow the case to be submitted to the jury. The dispute is wholly whether appellees have shown that a defect existed.
Under the doctrine of implied warranty, a defect is considered to exist in a product which is not fit for the ordinary purpose for which that product is used. Lonzrick v. Republic Steel Corp. (1966), 6 Ohio St. 2d 227. A manufacturer of a product impliedly warrants, in the sale of that product, that it is “of good and merchantable quality, fit and safe for * * * [its] ordinary intended use.” Lonzrick v. Republic Steel, supra, at 235. The claim made by appellees in the instant case is that the mountings used to attach the steering column to the instrument panel failed approximately 50 days after the automobile was purchased from Ford’s dealer, about nine months after the automobile was shipped from the factory. That claim is sufficient, if proved, to raise an issue that the mountings were not fit and safe for their ordinary and intended use, and were, therefore, defective.
*11The evidence at trial indicated that no one had serviced or tampered with the mounting assembly, which was original equipment on the automobile, at any time before the accident. This evidence was sufficient, if a defect were proved, to present a jury issue as to whether the defective condition existed at the time the product left the hands of the appellant.
The fundamental question which remains in the case is whether appellees have presented sufficient credible evidence to warrant submission to the jury of the issue of the existence of a defect in the mounting assembly, and (as a question of fact which is determinative) whether the steering column broke away while the automobile was still on the road, or whether it dropped, perhaps less than a second later, when the automobile struck the tree at a speed of at least 30 miles per hour, by the testimony of both appellees.
The trial court found that there was not sufficient credible evidence of a defect and sustained a directed verdict in favor of appellant.
The test to be applied to a motion for directed verdict was set out by this court in O’Day v. Webb (1972), 29 Ohio St. 2d 215:
“It is the duty of a trial court to submit an essential issue to the jury when there is sufficient evidence relating to that issue to permit reasonable minds to reach different conclusions on that issue, or, conversely, to withhold an essential issue from the jury when there is not sufficient evidence relating to that issue to permit reasonable minds to reach different conclusions on that issue.”
If there was sufficient evidence relating to the existence of a defect in the mounting assembly to permit reasonable minds to reach different conclusions, the case should have been submitted to the jury.
The evidence of appellees was their testimony that the steering column fell before the accident. Ordinarily, where testimony conflicts, the credibility of witnesses is a matter for the jury. However, in certain instances testimony can*12not be considered credible. Where a witness testifies that he looked and listened at a railroad crossing, but neither saw nor heard a train approaching, and the only reasonable conclusion upon the evidence is that there is no doubt that had he looked he must have seen the train, the witness’s testimony cannot be considered credible. Detroit, Toledo & Ironton Rd. Co. v. Rohrs (1926), 114 Ohio St. 493. See, also, Zuments v. B. & O. Rd. Co. (1971), 27 Ohio St. 2d 71.
The “railroad crossing” cases are a single example of the broad range of cases in which courts have recognized that eye-witnesses’ testimony, essential though it may be, is fundamentally “soft” evidence, subject to human failings of perception, memory, and rectitude. In law, as in other spheres of human affairs, simple facts may be far more persuasive than the most learned authorities. As in Dean Prosser’s homely example, “there is still no man who would not accept dog tracks in the mud against the sworn testimony of a hundred eye-witnesses that no dog has passed by.” Prosser on Torts (4 Ed.), 212.
The name generally given to this concept is the “physical facts rale.” The rule has been variously stated: E. g., “the testimony of a witness which is opposed to the laws of nature, or which is clearly in conflict with principles established by the laws of science, is of no probative value and a jury is not permitted to rest its verdict thereon.” Connor v. Jones (1945), 115 Ind. App. 660, 670, 59 N. E. 2d 577; Zollman v. Symington Wayne Corp. (1971), 438 F. 2d 28, 31-32, certiorari denied, 404 U. S. 827.
“The testimony of a witness which is positively contradicted by the physical facts cannot be given probative value by the court.” Lovas v. General Motors Corp. (1954), 212 F. 2d 805, 808.
“The palpable untruthfulness” of plaintiff’s testimony requiring a trial court to take a case from the jury under the physical facts rule “must be (1) inherent in the rejected testimony, so that it contradicts itself, or (2) irreconcilable rath facts of which, under recognized rules, the *13court takes judicial knowledge, or (3) is obviously inconsistent with, contradicted by, undisputed physical facts.” Duling v. Burnett (1938), 22 Tenn. App. 522, 124 S. W. 2d 294. Each of these formulations strikes a balance between, on the one hand, the common sense notion that physical facts and evidence can be so conclusive and demonstrative that no reasonable person could accept the truth of contrary testimony, and, on the other hand, the need for courts to be wary of treating a party’s theory of a case as “fact,” when a different theory is also possible in the case.
The physical facts rule has been applied in a myriad of cases and factual circumstances. Annotations 8 A. L. R. 796, 798; 62 A. L. R. 2d 1191, 1207. A few examples may be useful to show the applications of the rule:
In Galloway v. United States (C. A. 9, 1942), 130 F. 2d 467, affirmed, 319 U. S. 372, plaintiff sought to claim benefits under a war risk insurance policy issued by the federal government, claiming that service-connected causes had rendered bim totally and permanently disabled. The court held that the plaintiff’s two subsequent enlistments in the armed forces were such physical facts as to refute any reasonable inference from the evidence that he was totally and permanently disabled.
In automobile accident eases, the courts have looked to the particular facts of the cases to determine whether physical evidence conclusively favored one of the parties. See Black v. Berea (1941), 137 Ohio St. 611 (dictum). It is generally agreed that estimates of the speed or location of movable objects are not physical facts, so that in one case the testimony of a plaintiff that he traveled 30 feet at 20 miles per hour in the same period that defendant’s truck traveled 200 to 300 feet was not impossible, although this estimate would require the truck to have traveled at up to 200 miles per hour. The court held that the physical facts rule did not apply to estimates of time and speed. Hopfer v. Staudt (1956), 207 Ore. 487, 298 P. 2d 186. See, also, Raverty v. Goetz (1966), 82 S. D. 192, 143 N. W. 2d 859.
In these and other cases, the issue is generally wheth*14er circumstantial evidence of physical facts is so conclusive as to wholly rebut oral testimony presenting a different version. See, e. g., Black v. Berea, supra (137 Ohio St. 611). In a products liability ease, the physical evidence can assume a somewhat different role, since the physical fact of a defect is an element which must be proved by the plaintiff. Any claim of defect in a product is a claim in which the best direct evidence is the product itself.
In the instant case, the product in question is the assembly by which the steering column was mounted to the instrument panel. The steering column is about 2% feet long and is made up of two sections. The base is composed of a hard plastic formed into a wavy tube. Upon a sufficient impact, the base is designed to deform and collapse, absorbing some of the force which would otherwise be transmitted to the body of the driver. Above the base is a fixture, bolted to the steering column, which attaches to two brackets on the instrument panel. The direct attachment is by means of three aluminum blocks, each about an inch square and 3/8 of an inch thick. Each block has slots, about a quarter of an inch deep, along the length of two sides of the block. Two of these blocks are bolted onto brackets mounted on the instrument panel, the slots fitting into U-shaped metal flanges about an inch deep on the steering column fixture. The third block is bolted onto the fixture itself and fits a second bracket on the instrument panel. In assembling the automobile, the blocks are fitted into place in the U-shaped flanges, and plastic is then injected into the slots as sealant. Because the mountings of the blocks are open at the upper end, the blocks can break free of the plastic sealant if the steering column is pushed downward by more than an inch, as would happen upon an impact sufficient to collapse the base of the column to that degree. The column would, in such case, fall away from the instrument panel.
The steering column and the brackets from appellees’ vehicle show the signs of what happened in the accident. The base of the column is somewhat deformed. All of the aluminum blocks are intact; so too are the flanges in which the *15blocks were fitted, although these are slightly bent. On two of the mountings, pieces of plastic remain attached, with ragged edges showing where they were broken. Two other pieces of broken plastic sealant were found on the floor beneath the steering column. Photographs taken by both parties after the accident and the testimony of the witnesses show that all the bolts securing the mounting brackets and the steering column fixture were firm and tight after the accident. The inescapable conclusion from this physical evidence is that the steering column must have been compressed sufficiently so that the mounting blocks slid free, after which the steering column fell.
The testimony of appellant’s witnesses, as to the manufacturing specifications, and the actual physical condition of the steering column, confirm that an impact must have caused the breakaway of the steering column.
However, is this conclusion so certain that the physical facts rule should apply?
It is possible to suggest other ways in which a defect might have been responsible for the falling away of the steering column. The shock absorber might have been defective. Perhaps, in some manner an earlier minor blow had weakened the mounting. Perhaps, also, the connection between the fixture and the brackets was improper, in some manner which is not now apparent.
To raise these and other speculative explanations, however, does not make out a case that a defect in fact existed. This is so because they remain in the realm of speculation—there is simply no physical evidence to support any such hypothesis. Appllees did seek to suggest that a bolt was missing from the fixture; in fact, the uncontradicted testimony of appellant’s witnesses was that an empty hole in the fixture was not used with a bolt in this particular model of automobile. Appellees also presented testimony that the power steering was sometimes difficult to operate. No connection was suggested, however, between that claim and the existence of any defect in the mounting of the steering column.
The case, then, is starkly presented: On one side is the *16testimony of both appellees that something went wrong; on the other side is a theory that the mounting performed as it was intended. The former provides no explanation of the physical manner in which the falling away of the steering-column could have occured, and requires the assumption of events which are either contradicted by the physical evidence or for which there is no evidence at all; the latter provides an explanation of all the physical facts of the case, consistent with the proper and intended functioning of the mountings. It may be that some explanation other than Ford Motor Company’s is possible, but any such explanation would have to be based upon physical evidence which is wholly speculative. We find that appellees have failed to meet their burden of proof, and that reasonable minds could not differ, given the evidence submitted in this case, on the issue of the existence of a defect in the mounting of the steering- column.
This holding is in harmony with the decisions of other courts which have considered the application of the physical facts rule to the proof of defects in products liability cases.
In Zollman v. Symington Wayne Corp. (C. A. 7, 1971), 438 F. 2d 28, plaintiff claimed that an automobile hoist was defective, causing an automobile to fall from the hoist, pinning him underneath. Plaintiff claimed that he had properly positioned the front crossbar on the hoist against the tires. The defense was that a series of tests proved that the only way in which the accident could have happened was if the crossbar in fact had not been properly positioned. In tests, the automobile could not be made to fall when the- crossbar was properly placed, but would roll off if disturbed by a small amount of pressure when the crossbar was improperly positioned. In numerous tests performed by expert witnesses for either party, under various types of forceful assault, the vehicle fell only when the crossbar was improperly positioned. The court held that these tests and demonstrations were sufficient to render the plaintiff’s testimony of no probative value, and to entitle defendant to a directed verdict.
*17As in the instant case, one might speculate as to some other explanation than that provided by the physical tests, but such speculation, aided only by oral testimony, did not in Zollman raise sufficient doubt of an explanation based on physical tests of the normal functioning of the product, such that reasonable minds could find the presence of a defect.
In Lovas v. General Motors Corp. (C. A. 6, 1954), 212 F. 2d 805, the United States Court of Appeals considered a case arising in Ohio, involving facts very similar to the instant case. In Lovas, a tractor-trailer collided with a truck at an intersection, throwing the driver of the tractor from the cab and killing him. The tractor’s steering wheel was found lying close beside him. The administratrix of the deceased sued General Motors, the manufacturer of the tractor, for negligence in the construction or assembly of the steering shaft, claiming the driver had lost control because the steering wheel came loose. Witnesses testified that the screw threads on the nut and shaft were not damaged and that the steering nut was defective. The manufacturer contended that the force of the accident had driven the steering shaft downward, forcing the steering nut and the wheel off the shaft. The court noted that the actual nut and photographs of it taken after the accident showed no defect in the nut, and showed that threads on the nut and shaft were badly damaged, as they would be if forcibly pulled from the shaft. Other physical evidence also was in accord with the manufacturer’s theory. The court held that the sustaining of a directed verdict was proper, stating that “In order to have her ease submitted to the jury, appellant had the burden of introducing evidence of probative value which would tend to show that the steering wheel came off the steering shaft because of the negligent construction or assembly of the wheel and shaft, and that the accident was caused by such a defect, ’ ’ and that ‘ ‘ The testimony of a witness which. is positively contradicted by the physical facts cannot be given probative value by the court.” Lovas v. General Motors Corp., supra, at 807, 808.
Any case in which the trial judge decides that there is *18not sufficient credible evidence to support an element of a litigant’s proof must be controlled by the particular facts involved. In such cases, courts must take care, recognizing that unlooked-for consequences are as fully a part of human experience as consequences both expected and predicted. Yet, any finding by a jury must be based on credible evidence. In the instant case, any jury verdict for the appellees would of necessity be grounded on sheer speculation as to the existence of facts for which no evidence was presented. Appellees’ oral testimony, unsupported, is not sufficiently credible to overcome the evidence of physical facts which are compatible only with the theory that there was no defect in the mounting of the steering column. At the trial of this case, the trial judge carefully considered the evidence, and we agree that there was not sufficient evidence to submit this case to the jury.
The judgment of the Court of Appeals is reversed.
Judgment reversed.
0’Neill, C. J., Corrigan, W. Brown and P. Brown, JJ., concur. Herbert, J., concurs in the judgment. Celebrezze, J., dissents.