On July 7, 1967, Maurizio Fortunato drove his six weeks old 1967 Ford Mustang from his home in Weehawken, New Jersey, to a dance hall in Manhattan. After arriving there about 10:00 or 10:30 p. m., he met a girl with whom he danced and drank two or three highballs. He left about an hour later and took the girl to her home in Queens. From there he headed for Manhattan by way of the Queensboro Bridge. Just as he was approaching an “S” curve on the ramp leading from the upper level of the bridge, the inside of his car erupted into flames as he was lighting a cigarette with the car’s lighter. The car then went out of control, struck the right side of the bridge, and caromed to a stop astride the divider in the left lane. He was pulled from the burning car by a tow truck operator, Joseph Comparato, who had been traveling behind Fortunato on the bridge. The plaintiff had suffered second and third degree burns over 45% of his body. An ambulance was called and he was rushed to a hospital.
At the trial, the principal issue was whether or not any defect in the design or manufacture of- the Ford Mustang was the cause of Fortunato’s injuries. The plaintiff’s theory was that the Mustang gas tank design was unsafe because the tank formed an integral part of the body of the car, i. e. the top of the tank served also as the floor of the trunk, instead of being separately strapped onto the outside of the car with a space between the top of the tank and the floor of the trunk as it is in most other makes of automobiles. In addition, the plaintiff offered proof that an imperfection in the metal used for the top of the tank gave rise to a small hole which allowed gasoline vapors to escape into the trunk of the car and eventually to diffuse into the passenger compartment, where they were ignited by the cigarette lighter. Fortunato’s claims were corroborated by Comparato who testified that he saw the car burst into flame from the inside before it hit the bridge wall and also by several expert witnesses who testified that the combination of factors described above was the probable cause of the accident.
Ford, in turn, argued that any hole in the gas tank was caused by rust after the accident and not by any fault in the metal. But as its most important point, Ford maintained that Fortunato’s explanation of the accident was physically impossible and that the accident hap*965pened because Fortunato, who had been drinking and was unfamiliar with the bridge, failed to negotiate the curve and caused the car to catch fire by rupturing the gas tank against the bridge wall.
In support of its view that the plaintiff’s claim was untenable, Ford introduced the testimony of one of its employees, Paul Toth, who had run some tests on another 1967 Mustang in Dear-born, Michigan, during September of 1971. Toth testified that he punched a hole in the gas tank just slightly larger than the one in Fortunato’s tank and then drove the car various distances under differing conditions, including some with the windows open and some with them shut. At no time was he able to create a heavy enough gasoline vapor mixture to be explosive; he said, however, that whenever there were fumes in the passenger area, they gave off a strong odor. Toth did testify that after leaving the car standing in the sun with the windows closed for ten minutes, the vapor mixture in the car was 86% of the explosion point, and he estimated that it would have been reached in another five minutes.
Ford claims that it was entitled to a directed verdict or a judgment notwithstanding the verdict on the basis of Toth’s tests which proved that Fortunato’s car could not have caught on fire in the way he explained, especially in light of the fact that Fortunato never smelled any gasoline odors and he was driving at night with the driver’s window half open.
I
On this appeal, we need not determine whether or not the federal or the state sufficiency of the evidence standard is to be applied because the results would be the same under either test, cf. O’Connor v. Pennsylvania R. Co., 308 F.2d 911, 914 (2 Cir. 1962). In determining whether the motions for a directed verdict and a judgment n. o. v. were properly denied, this court must “view the evidence in the light most favorable to [Fortunato] and give [him] the benefit of all inferences which the evidence fairly supports, even though contrary inferences might reasonably be drawn,” Continental Ore Co. v. Union Carbide & Carbon Corp., 370 U.S. 690, 696, 82 S.Ct. 1404, 1409, 8 L.Ed.2d 777 (1962); see also Simblest v. Maynard, 427 F.2d 1, 4 (2 Cir. 1970). Taking that view of the evidence we must then decide whether or not “reasonable men” could return a verdict for the plaintiff, see O’Connor, supra, 308 F.2d at 914-915; 5A J. Moore, Federal Practice, [[50.02 [1], at 2320 (2d ed. 1971).
Ford, of course, does not dispute the fact that there was more than sufficient evidence introduced by the plaintiff to support a verdict on his behalf, but it argues, rather, that much of that evidence must be disregarded because it is contrary to proven physical facts. This so-called “physical facts rule” has been widely adopted, but its application to specific cases calls for some care as it is a determination by the court that certain facts may not be found by the jury. Such a decision should not be lightly made because unexpected and quite unusual factual situations have often occurred, and the duty of the court is not to decide whether the defendant’s theory is probably right, but to determine whether or not an element of the plaintiff’s case, which it claims to have proved and upon which a verdict for the plaintiff would have to be based, is physically and factually impossible, see Born v. Osendorf, 329 F.2d 669, 672 (8 Cir. 1964). As Professor James has pointed out, the doctrine of excluding evidence because it is incredible as a matter of law is used sparingly, F. James, Civil Procedure, § 7.11, at 273-74 (1965), and the trend is toward even greater restraint in its use on the part of the courts, 2 F. Harper & F. James, The Law of Torts, § 15.2, at 877 (1956).
When applying the doctrine the courts have been quick to indicate that the physical fact relied upon must not itself be in doubt by saying that it must be *966“uncontested”, O’Connor, supra, 308 F.2d at 915; “incontrovertible”, Dostal v. Baltimore & O. R. Co., 189 F.2d 352, 354 (3 Cir. 1951); “uneontradicted”, Granat v. Schoepski, 272 F.2d 814, 815 (9 Cir. 1959); “undisputed”, Zollman v. Symington Wayne Corp,, 438 F.2d 28, 31 (7 Cir.), cert. denied, 404 U.S. 827, 92 S.Ct. 59, 30 L.Ed.2d 55 (1971); or “clearly contrary to an immutable law of physics,” Kansas City Public Service Co. v. Shephard, 184 F.2d 945, 948 (10 Cir. 1950).
Proof of this standard has been supplied when the physical fact was admitted by the opposing party, Zollman, supra; Peretore v. Pennsylvania R. Co., 284 App.Div. 472, 131 N.Y.S.2d 696, 697 (1954), was an easily measured distance or height; Hunter v. New York, O. & W. R. Co., 116 N.Y. 615, 23 N.E. 9 (1889); Moore v. Bd. of Education, 22 A.D.2d 919, 255 N.Y.S.2d 540 (1964) aff’d 19 N.Y.2d 621, 278 N.Y.S.2d 406, 224 N.E.2d 879 (1967); or was a weather report supported by neutral documentation, O’Connor, supra.
For a number of reasons, the tests performed by Toth do not bring the case within the physical facts rule. For example, we know nothing of the condition of the test car other than that it was a 1967 Mustang. The distances driven, the type and amount of gasoline in the tank, the comparative smoothness or roughness of the roads, the amounts of resulting sloshing of gasoline in the tanks and the relative amounts of gasoline fumes which may have been thereby forced through the holes are not shown to have been approximately the same. All the test measurements were made by using a combustibile gas indicator, but it is not known whether these measurements were made at floor level or seat level where the fumes would be more dense or halfway from floor to ceiling or at ceiling level where they would be less dense, nor was a test made with a cigarette lighter at different levels in a closed chamber, comparable to that of a car in atmospheric conditions of pressure and temperature like those at the time of the accident.
In addition, all of the tests were run by an employee of the defendant who was asked to make them in connection with this case. They were not subject to any independent verification. Furthermore, no written report on the outcome of the tests was made by Toth prior to the trial. Test results should not even be admissible as evidence, unless made by a qualified, independent expert or unless the opposing party has the opportunity to participate in the test, C. McCormick, Evidence, § 169, at 362 (1954). Certainly the jury should be permitted to pass on the credibility of the witness and the accuracy with which he conducted his experiments, cf. Jarman v. Philadelphia-Detroit Lines, 131 F.2d 728, 730 (4 Cir. 1942). Though the purpose of the tests was to demonstrate the impossibility of plaintiff’s theory, they did show that under certain circumstances, very high quantities of gasoline vapor did enter the passenger compartment of the car.
The plaintiff’s allegation was supported, not only by his own testimony that the fire came from inside the ear but by that of the neutral eyewitness Comparato; and in contradiction to Toth’s experiments was expert testimony that such an occurrence was a distinct possibility. Considering the mass of reputable conflicting evidence, this case was properly submitted to the jury and was clearly not a case in which to apply the physical facts rule, cf. Russell v. Gonyer, 264 F.2d 761 (1 Cir. 1959); Dostal, supra, 189 F.2d at 355.
Therefore, even if the tests and their results are assumed to be strong evidence in favor of Ford’s position, they fell far short of the uncontroverted proof required to justify a court in taking the issue from the jury. The denial of the motions for a directed verdict and a judgment n. o. v. are affirmed.
*967II
Ford argues that a new trial is required because four items of evidence were excluded. In no instance, however, did Ford make an offer of proof, as provided for by Fed.R.Civ.Proc. 43(c), and it is therefore necessary to make some general comments on this failure before looking at the specific allegations of error.
While Rule 43 does not mandate an offer of proof, and this court has not held that such an offer is an absolute prerequisite in every appeal from the exclusion of evidence, see, e, g., Massachusetts Mut. Life Ins. Co. v. Brei, 311 F.2d 463, 465 n. 1 (2 Cir. 1962); Meaney v. United States, 112 F.2d 538, 539 (2 Cir. 1940), it is required where the significance of the excluded evidence is not obvious or where it is not clear what the testimony of the witness would have been or that he was even qualified to give any testimony at all, Herencia v. Guzman, 219 U.S. 44, 46, 31 S.Ct. 135, 55 L.Ed. 81 (1910); Marrone v. United States, 355 F.2d 238, 241 (2 Cir. 1966); Moss v. Hornig, 314 F.2d 89, 93 (2 Cir. 1963); 5 J. Moore, Federal Practice, H 43.11, at 1381-1382 (2d ed. 1971).
The main purposes for this rule are to permit the trial judge to reevaluate his decision in light of the actual evidence to be offered, c/., McCormick, Evidence, § 51, at 113-14 (1954), and to permit the reviewing court to determine if the exclusion affected the substantial rights of the party offering it, Fed.R.Civ.Proc. 61. An appellate court will not reverse on the mere possibility that the exclusion was harmful, Palmer v. Hoffman, 318 U.S. 109, 116, 63 S.Ct. 477, 87 L.Ed. 645 (1943); Boulter v. Chesapeake & O. R. Co., 442 F.2d 335, 336 (6 Cir. 1971); Sorrels v. Alexander, 79 U.S.App.D.C. 112, 142 F.2d 769, 770 (1944), nor will it permit a party to allege on appeal what it failed to claim to the trial court. To do so would allow a party to obtain a new trial simply on its claim that it would have proven a certain fact or facts had it been given a chance.
More specifically, Ford first asserts that it was improperly barred from showing that out of a large number of cars sold no owner had complained of a defect similar to that claimed by Fortunato. Even assuming that such evidence would be admissible,1 it is clear that there was no error in the exclusion of it here. Ford’s only attempt to show the absence of complaints was through the cross-examination of two of Fortunato’s expert witnesses, Edward Gloyd and Daniel O’Connell. Gloyd, who was employed in the Safety Office of the Chief of Engineers of the United States Army, was asked, “Do you know how many of the Ford line the Army buys a year?” Objection to the question was sustained. Counsel for Ford did not explain to the trial court what the purpose of the question was or what line of inquiry he was seeking to pursue, nor did he ever make an offer of proof. O’Connell, a licensed consulting engineer in New York, was asked if he had any idea of the number of Mustangs made and sold in the years 1964 to 1967. Again objection was sustained; no explanation of the purpose was given, and no proper foundation was laid.
It is difficult for us to see what relevance these questions, asked of witnesses not connected with the Ford *968Motor Company, could have had in demonstrating a lack of prior complaints to Ford concerning the gas tank design of a 1967 Mustang. The court was not in error in excluding answers to these questions, Herencia, supra; Marrone, supra.
Ford next complains that it was not permitted to show the advantages of its gas tank design in order to counter Fortunato’s claim that it was unsafe; however, Ford never made any attempt to do so. What it did try to do was to ascertain through one of its design engineers, Gilford Gorker, whether the thickness of the metal in the Mustang tank was greater than that used in the more customary, strapped-on tanks. Even if we assume that Gorker would have testified that the Mustang tank was thicker, this evidence would not be relevant to this case in which the issue was whether or not there was a hole in the tank which allowed fumes to go into the passenger compartment of the car. If there were such a hole, it would not matter how thick the tank was, and the claimed design defect was that there was no space between the top of the gas tank and the floor of the trunk.
Ford’s third contention is that it was unable to show the dangerous nature of the "S" curve on the bridge, which would have supported its theory of the accident. Competent evidence of accidents similar to Fortunato’s on that segment of the bridge would have been admissible, see, e. g. Dist. of Columbia v. Armes, 107 U.S. 519, 524-525, 2 S.Ct. 840, 27 L.Ed. 618 (1883); Kaplan v. City of New York, 10 A.D.2d 319, 200 N.Y.S.2d 261 (1960); McCormick, Evidence, § 167, at 350-51 (1954); however, the exclusion complained of was not error.
Ford had called Thomas McNulty, the patrolman who investigated the Fortunato accident, as a witness and had elicited from him, without objection, the fact that he had been called to two or three other accidents on the bridge prior to the one in question here. The defendant then asked McNulty if there was a high accident rate chart in the station house, but an objection was sustained to the question.
The question was clearly improper because any information contained on the chart would be hearsay as testimony of McNulty. Although the records forming the basis of any such chart, and perhaps the chart itself, would be admissible under the Federal Business Records Act, 28 U.S.C. § 1732(a), there was no attempt to enter the records nor was any effort made to lay a proper foundation for offering them, cf., United States v. Dawson, 400 F.2d 194, 198-199 (2 Cir. 1968), cert. denied, 393 U.S. 1023, 89 S.Ct. 632, 21 L.Ed.2d 567 (1969). Therefore, this court has no means of knowing whether or not the “S” curve was a high accident area and whether or not any other accidents followed a pattern similar to that in the present case.
Finally, Ford claims error in the exclusion of testimony designed to impeach Comparato by showing a prior inconsistent statement. Comparato had testified that he saw Fortunato’s car erupt with flames from the inside before it struck the bridge, but Ford claims that Comparato had previously told one of its investigators, John Hughes, that he first saw flames outside the Mustang before it hit the bridge and then flames also appeared inside after the car had hit the retaining wall and the divider. Hughes died, however, prior to trial, and Ford attempted to get in the impeaching testimony through Donald Conway, who testified that he was present when Hughes made a telephone .call to Comparato.2 The transcript does not re*969veal whether Conway was on an extension line 3 or if he just heard Hughes’ end of the conversation, but it does reflect the fact that Conway only knew that Hughes was talking with Comparato because Hughes told Conway that he wtas telephoning him.
During Ford’s cross-examination of Comparato, the latter acknowledged that he had spoken to a man on the telephone who might have been Hughes; however, Comparato claimed that he told the person on the telephone that he would not answer any questions concerning the accident.4 Although it is the rule that a prior inconsistent statement cannot be used to discredit a witness unless he has first been asked whether or not he made such a statement, United States v. Hayutin, 398 F. 2d 944, 953 (2 Cir.), cert. denied 393 U. S. 961, 89 S.Ct. 400, 21 L.Ed.2d 374 (1968), Comparato was never asked.
The court cannot accept Ford’s self-serving statement in its brief on appeal of what the alleged inconsistent statement was. The purpose of the line of inquiry and an offer of proof must be made at the trial where the witness is available and the trial court is in a position to reconsider its decision.
III.
Ford also contends that the plaintiff’s summation, in a number of respects, was so unfair as to require a new trial; however, an examination of the closing arguments of counsel in the light of the transcript and other parts of the record makes clear that this claim is without merit.
Error is claimed in relation to counsel’s argument concerning the cutout piece of metal containing the small hole which, Fortunato alleged, permitted gasoline vapors to escape from the gas tank. There was testimony at the trial that the plaintiff’s expert metallurgist, Dr. Morfopoulos, had cut this piece from the tank in order to make photographs and give it close scrutiny under a microscope. Dr. Morfopoulos also testified that he had returned the piece of metal to plaintiff’s counsel and had heard that it had been lost. The rest of the gas tank was put into evidence, but the piece *970with the actual hole in it was never produced. There were in evidence, however, numerous pictures of the hole from several angles and at various magnifications.
During argument plaintiff’s counsel told the jury that he did not know where the missing piece was and asked the jury to “throw us out of court” if it felt he was at fault for losing the piece of metal.5 These remarks were improper as an effort by counsel to testify, in effect, that he did not know where the metal was; but the court immediately instructed the jury to disregard the statements of counsel that were not supported by facts in evidence and under the circumstances we conclude that the error was not so substantial as to warrant an order for a new trial. In his charge, the judge correctly instructed the jury that it might draw a strong inference against the plaintiff, if it determined that the failure to produce the piece of metal was intentional,6 cf., Matarese v. Moore-McCormack Lines, 158 F.2d 631, 637 (2 Cir. 1946); Austerberry v. United States, 169 F.2d 583, 593 (6 Cir. 1948); 2 J. Wigmore, Evidence, § 291, at 180-188 (3d ed. 1940).
Furthermore, Ford has never claimed any actual prejudice from its failure to have physical access to the metal. Upon discovering that the metal was missing, a consent order was entered permitting Ford to take the deposition of Dr. Morfopoulos, and its own expert metallurgist, Dr. Eisenberg, testifying from the photographs, stated categorically, at least seven times, that the hole was caused by rust after the accident and not by an imperfection in the manufacturing process. Not once did Dr. Eisenberg indicate that his opinion was hindered by his inability to see the actual hole, even though defense counsel made *971repeated reference to the fact that the metal was missing.
Ford claims as other grounds for a new trial that counsel for the plaintiff made a number of prejudicial and unfair arguments to the jury in his closing summation. It asserts that he played upon the sympathies of the jurors and that he called upon the jury to punish Ford by its verdict. The court specifically charged the jury that it could not award punitive damages and, considering the severity and extent of the injuries, this court cannot say that the verdict, though high, was excessive or resulted from sympathy or an intent to punish Ford.7 The appellant’s remaining points are minor and not of sufficient merit to call for discussion.
The judgment of the district court is affirmed.
. It is highly questionable -whether evidence showing the absence of any similar complaints would be admissible in this case, which was primarily one of a breach of warranty, because the plaintiff withdrew any reliance on a negligence theory during trial. A breach of warranty gives rise to strict liability which is not necessarily negated by a showing of due care or a lack of knowledge of the defect on the part of the manufacturer or seller, Blessington v. McCrory Stores Corp., 305 N.Y. 140, 111 N.E.24 421 (1953); W. Prosser, The Law of Torts, § 95 at 652 (3d ed. 1964). Therefore, evidence of a lack of prior complaints is rarely probative on the warranty issue, see Frank R. Jelleff Inc. v. Braden, 98 U.S.App.D.C. 180, 233 F.2d 671, 677-80 (1956).
. The direct examination of Conway included the following:
“Q. Do you have any knowledge as to what Mr. Capirado [sic] said to Mr. Hughes ?
Mr. Peters : Objection to this.
The Court: Mr. Conway, it’s so clearly hearsay unless you are going to have some reason why Mr. Conway could *969speak about this of his own knowledge, I don’t see how I could possibly—
Q. Mr. Hughes is now dead?
A. Yes.
Q. When did be die? A. August 14, 1971.
Q. Did you and Mr. Hughes share an office? A. We did.
Q. Were you present when a telephone conversation was had between Mr. Hughes and Mr. Capirado? [sic]
Mr. Peters: I object to that when a conversation was had between somebody else and somebody else.
The Court: He can say he was present.
Mr. Peters: How does he know who was on the other end of the phone?
The Court: Would you have a basis of knowing?
The Witness: He told me he was calling him.
The Court: Again I believe that is hearsay.
Mr. Conway: All right, sir.
You never spoke to Mr. Capirado [sic] yourself?
The Witness: No, I did not.
Mr. Conway: I have no further questions.
Mr. Peters: I have no questions, your Honor.”
. We need not decide the difficult question of whether or not circumstantial evidence would have been sufficient to identify Comparato to Conway as the person on the other end of the line assuming that Conway did hear that voice, cf. United States v. Fassoulis, 445 F.2d 13, 17 (2 Cir.), cert. denied, 404 U.S. 858, 92 S.Ct. 110, 30 L.Ed.2d 100 (1971) ; United States v. Benjamin, 328 F.2d 854, 861 n. 3 (2 Cir.), cert. denied, Howard v. United States, 377 U.S. 953, 84 S.Ct. 1631, 12 L.Ed.2d 497 (1964), because there was no proper foundation laid for the impeachment.
. The cross-examination of Comparato, with reference to the alleged inconsistent statement, included only the following:
“Q. Didn’t he [Hughes] have a long telephone conversation with you reviewing what you had said before? A. No, I told him I wouldn’t answer any questions, only to tell one story.”
. Summation by Mr. Peters for the plaintiff included the following:
“ . . . Take a look at the size of the piece that is missing. It is half the size of an old tliree-eent stamp, and that piece was sectioned in several pieces so it could be cross-photographed in a microscope.
Do you know what we are dealing with in our office? Specks. I can’t tell you where it is, if I could I wish to God I would. I don’t know where .it is. I never even—
Mr. Conway: Your Honor, I am going to object to Mr. Peters trying to get some testimony in at this point.
The Court: I think enough has been said on that, Mr. Peters.
Mr. Peters: The court is going to charge you that if we intentionally lost that piece so as to prevent you or them from seeing it, you may draw the greatest of inference from that, and I tell you right now I want you to go further than the court charges you against me if you feel that I or anybody from my office intentionally lost that piece—
Mr. Conway: This is objected to, there is no such testimony, and now Mr. Peters—
Mr. Peters: May I continue?
Mr. Conway: Is he going to do some testifying?
The Court: Please, Mr. Peters, let’s get beyond that.
Mr. Peters: I say throw us out of court, if it is my fault or anybody’s fault, regardless, throw us—
Mr. Conway: I object to this. There is no proof on this one way or the other except Mr. Peters’ unsupported, unsworn statement that he is now trying to make and I object and I insist, Your Honor, it is improper.
The Court: The jury will be told again to disregard statements of counsel with respect to whether or not the facts they ask you to draw are facts.”
. The court charged :
“There is one further matter for your determination with respect to the evidence. A small piece of the gasoline tank in evidence was not produced by the plaintiff whose attorneys last had it in their control.
This piece, it has been testified, contains the tiny hole, the existence, nature, cause and duration of which is one of the sharply disputed and important issues in this case. It is for you to determine whether or not under all the surrounding circumstances and evidence the failure to produce this piece of physical evidence was intentional. If you so find you may draw a strong inference that its production in the trial would have been unfavorable to the plaintiff’s interest.”
. Ford does not claim that the $400,000 verdict was excessive. Fortunato had special medical damages of $50,000, lost earnings of $15,000, and a potential future loss of wages exceeding $150,000, in addition to suffering intense pain for many months after the accident and permanent physical and psychological injuries.