JOHN W. PECK, Circuit Judge.

The plaintiff-appellant, Crown Cork & Seal Company, Inc. (hereinafter referred to as Crown) originally filed suit in the District Court for the price of goods sold to the defendant-appellee, Morton Pharmaceuticals, Inc. (hereinafter referred to as Morton). Morton counterclaimed for damages, alleging that some 36,000 aerosol cans previously purchased by it from Crown were defective. The District Court granted Crown’s motion for summary judgment on its claim; the issues arising out of the counterclaim were tried to a jury. This appeal is from the judgment of the District Court entered on the jury verdict in Morton’s favor on the counterclaim.

As its name implies, Morton, a Tennessee corporation, is in the business of manufacturing and distributing various pharmaceutical products including aerosol spray products. Crown, a New York corporation, is the world’s largest manufacturer of aerosol cans.

*923 In October, 1962, Morton received an order from Globe Laboratories, Inc., of Fort Worth, Texas, to prepare 36,000 cans of an aerosol spray called “Kleen Sheen,” a pet grooming aid. For many years Morton had been purchasing empty aerosol cans from Crown, and therefore it ordered 36,000 empty, pre-lithographed aerosol cans from Crown. Upon receipt of the empty cans from Crown, Morton filled them, inserted the valve assembly and shipped the filled cans to Globe. Shortly after Globe received the shipment, cans began leaking. At that time Globe sent approximately 144 cans back to Morton. Morton immediately notified Crown of the problem, and Crown sent a representative to Morton’s plant to investigate.

While the results of this investigation and the cause of the original leakage are in dispute, it is undisputed that at the time of the first investigation, the parties believed that all of the cans that were going to leak had leaked, and therefore the parties advised Globe that it could go ahead and distribute the rest of the “Kleen Sheen” cans. However, shortly after distribution by Globe, many more “Kleen Sheen” cans began to leak and Globe recalled all of the “Kleen Sheen” cans.

The theory of Morton’s counterclaim was that the cans were defective because negligently manufactured and that the defects violated the statutorily implied warranties of merchantability and fitness for the purpose intended. 1

It should be noted at the outset that there is no dispute that approximately three thousand of the 36,000 cans did in fact leak. Nor is there any dispute that leakage in aerosol cans is not necessarily due to manufacturing defects in the cans since improper filling or insertion of the valve assembly can also cause can leakage. Thus the primary issue in the trial court was whether the admitted can leakage was caused by defects in the cans, and the primary issue here is whether sufficient evidence was presented in the trial court that the leakage was caused by can defects, as the jury necessarily found.

It is Crown’s contention here that there was no substantial evidence that the leakage was caused by defects in the cans, and that therefore the District Court erred in refusing to grant Crown’s motions for a directed verdict made at the close of the plaintiff’s evidence and renewed at the close of all the evidence. In the alternative, Crown contends that the District Court abused its discretion by denying its motion for a new trial on the ground that the verdict was against the weight of the evidence. We cannot agree with these contentions.

Mr. Gordon Chissom, a former employee of Crown, testified with respect to his examination of the first group of 144 cans which were returned to Morton by Globe as follows:

«Q. * * * [H]ow would you classify the cans on questionable manufacturing ?

A. The cans that I examined I said had a questionable seam. In terminology we call them a tight seam, or semi-tight or loose seam. This is a bottom seam.

Q. If you characterize one as having a loose seam is it likely to leak?

A. It could or it could not.

Q. Did you find loose seams in these cans?

*924 A. I did. Loose bottom seams. Let’s qualify it.”

Mr. Wayne Speer, a former employee of Morton, testified that he and Gordon Chissom jointly examined leaking cans out of the first group of 144 cans returned to Morton by Globe. Speer testified that Chissom told him that he had found a number of difficulties in the cans in the soldering of the side seams of the cans, at the “crossovers” where there are two layers of metal on the cans, and that the openings or “beads” of the cans were elliptical or “cracked” rather than round. 2

Crown vigorously stresses that both of the above items of testimony relate to the first group of cans which Morton received back from Globe. At the time, both Crown and Morton believed that the defects which caused the leakage in the first group of returned cans were due to normal manufacturing defects, and that when the leaking cans were removed from the entire shipment there would be no further leakage problem. However, as recited above, a much more serious leakage problem developed in the “Kleen Sheen” cans after Globe distributed the cans to its customers for retail sale. Crown contends that none of this later leakage was related to any manufacturing defects which may have been found in the 144 cans which were originally returned to Morton.

However, Speer also testified that the second group of cans, the entire shipment which Globe returned to Morton, contained cans which were leaking at the side seams, around the opening at the top, and at the bottom seams. While the terminology of the witness was that of “leakage” rather than “defects” in the cans, the witness was describing the same general type of problem with the second group of returned cans as was found with the first group of 144 cans. The jury could reasonably infer that the leakage described with respect to the second group of cans was a result of the defects attributed to the first group.

In addition, Speer testified that Gordon Chissom informed him that the results of tests performed by Crown in their own laboratory showed that at least some of the leakage was attributable to a deficiency in the compounding of the valve cup inserted by Morton into the empty cans purchased from Crown. Morton purchased the valve cups from the Precision Valve Corporation, but Speer testified that Chissom informed him that Morton did not have to include Precision Valve Corporation in the can leakage claim because Crown manufactured a very large percentage of the valve cups for the Precision Valve Corporation. 3 Chissom corroborated Speer’s testimony on this point.

*925 Finally, there was evidence that following the leakage problem with the cans in controversy, Morton purchased additional empty cans of the same type and size from Crown, filled them with the same “Kleen Sheen” formula, used the same method of inserting the valve assembly, and packaged and shipped the filled cans in the same manner as was done with the cans in controversy. There was no leakage problem with these later cans.

Thus, although Crown introduced much persuasive expert testimony and other evidence which would tend to show that the cause of the leakage was improper packaging of the aerosol product in the cans by Morton and not manufacturing defects in the cans, viewing the evidence most favorably to Morton, we cannot say that reasonable minds could only reach one conclusion, namely, that the leakage was caused by factors other than manufacturing defects. See Brady v. Southern Railway Company, 320 U.S. 476, 64 S.Ct. 232, 88 L.Ed. 239 (1943); Delk v. St. Louis and San Francisco Railway Company, 220 U.S. 580, 31 S.Ct. 617, 55 L.Ed. 590 (1911); Pearl Assurance Company v. Stacey Brothers Gas Construction Company, 114 F.2d 702 (6th Cir.1940); 5 J. Moore, Federal Practice ¶ 50.02 [1] (2d ed.1968) and cases cited therein. Nor can we say that the District Court clearly abused its discretion by denying Crown’s motion for a new trial on the ground that the verdict was against the weight of the evidence. See Duncan v. Duncan, 377 F.2d 49 (6th Cir.1967); Cutter v. Cincinnati Union Terminal Co., 361 F.2d 637 (6th Cir.1966); Southern Railway Company v. Miller, 285 F.2d 202 (6th Cir.1960); 6A J. Moore, Federal Practice ¶ 59.08[5] (2d ed.1966). We therefore hold that the District Court correctly submitted to the jury the question as to whether the leakage was due to defects in the cans, and did not abuse its discretion by refusing to overturn the jury’s verdict on the weight of the evidence.

The next issue concerns the communication (and therefore the admissibility) of an express warranty and limitation of liability provision appearing on the back of the standard form of order acknowledgment used by Crown. Crown made an offer of proof which showed that in response to every order received by it, a master order set was prepared. The master order set consisted of a pre-pack-aged set of forms on which the necessary information about the order could be typed just once, making numerous copies for use in the production, scheduling, shipping and invoicing of the order. Also contained in the master set was the order acknowledgment, and it was the only form in the set to contain the limitation of liability provision. The offer of proof further showed that it was Crown’s unvarying procedure to send the acknowledgment copy containing the express warranty and limitation of liability to every customer on every order.

The District Court refused to admit the proffered evidence, which consisted of a blank copy of a standard form acknowledgment containing the limitation of liability provision and the testimony outlined above concerning Crown’s usual procedure in handling orders and sending acknowledgments, because Crown could produce no witness who could testify that an acknowledgment was prepared and mailed for the order for the cans in controversy, while Morton’s president testified that it did not receive an acknowledgment for the order for the cans. The District Court’s ruling was clearly correct. If evidence that an acknowledgment was prepared and mailed for the cans in controversy *926 had been offered, the jury could have properly inferred that Morton received such acknowledgment. Here, however, Crown was asking the Court to permit the jury to infer first, that an acknowledgment was prepared and mailed for this particular order because one was normally prepared and mailed in the ordinary course of business, and second, that Morton received such acknowledgment. As the District Court correctly pointed out, this would require an impermissible inference that the acknowledgment was received by Morton to be based on an inference that it was sent. The rule that an inference cannot be based on an inference is well established in Tennessee. See, e.g., Western and Atlantic Railroad v. Land, 187 Tenn. 533, 542-543, 216 S.W.2d 27, 31 (1948); Edenton v. McKelvey, 186 Tenn. 655, 660, 212 S.W.2d 616, 618 (1948); Shockley v. Morristown Produce and Ice Company, 158 Tenn. 148, 161, 11 S.W.2d 900, 904 (1928); East Tennessee and Western North Carolina Railroad Company v. Lindamood, 111 Tenn. 457, 473, 78 S.W. 99, 103 (1903). We therefore hold that the District Court committed no error in excluding the proffered evidence concerning the limitation of liability provision contained in the order acknowledg- - ment form.

The final issue which we consider is whether the District Court’s exclusion from evidence of the results of certain tests on the cans in controversy constituted prejudicial error. As outlined above in the recital of facts, Morton filled the empty cans purchased from Crown with the aerosol product and inserted the valve cup assembly in the can. The process of sealing the valve cup assembly with the can is called “clinching” or “crimping” the can. The evidence proffered by Crown consisted of the results of tests on the cans in controversy in which a newly developed micrometer was used to measure the depth of the crimp around the valve cup assembly. Crown contends that the evidence would show that the depth of the crimp was improper, resulting in a bad seal between the can itself and the valve cup assembly, and further that such improper crimping was the primary cause of the leakage.

The District Court ruled that since the micrometer had not been invented at the time the cans were filled and crimped by Morton, Morton could not be held to the standard of performance shown to be possible by the new device. This ruling was clearly erroneous. The issue was not whether Morton should be held liable for failure to know of or use such a measuring device, but whether the results of the tests would be probative of and relevant to one of the primary issues arising at the trial, the cause of the can leakage. Experimental evidence is admissible so long as the evidence is relevant and probative, and experimental evidence is deemed to have probative value if the conditions of the experiment were identical with or similar to the conditions of the transaction in litigation. See Saldania v. Atchison, Topeka and Santa Fe Railway Company, 241 F.2d 321, 322 (7th Cir.1957); Lever Brothers Company v. Atlas Assurance Company, 131 F.2d 770, 777 (7th Cir. 1942); Mauldin v. Otto Schwill and Company, 1 Tenn.App. 347, 354 (1925). Since one of Crown’s primary defenses was that the leakage was caused by faulty crimping of the cans, a factor under Morton’s control, and since the tests were conducted on a sample of the actual cans in controversy, the results of the tests would certainly have been relevant and had probative value.

There remains, however, the question of whether the exclusion of the evidence was prejudicial to Crown. The erroneous exclusion of evidence in a jury trial is harmless only if no substantial rights of a party have been violated. Segal v. Cook, 329 F.2d 278, 280 (6th Cir.1964); 28 U.S.C. § 2111 (1964); Fed.R.Civ.P. 61. With respect to the erroneous exclusion of evidence in a jury trial, it has often been held that the exclusion is not prejudicial if other substantially equivalent evidence of the *927 same facts has otherwise been admitted into evidence. See, e.g., United States for Use and Benefit of Carter v. Ross Corporation, 385 F.2d 564, 566 (6th Cir.1967); United States v. Stoppelmann, 266 F.2d 13, 20 (8th Cir.1959); MacDonald v. Milwaukee Mechanics’ Insurance Company, 167 F.2d 276, 278 (7th Cir.1948), cert. den., 334 U.S. 845, 68 S.Ct. 1514, 92 L.Ed. 1769 (1948). While Crown successfully introduced expert opinion evidence tending to show that the cause of the leakage was faulty crimping of the valve cup assembly to the can, such expert opinion evidence cannot be considered the equivalent of factual data based on the results of scientific tests, particularly in this case where the proof of both parties consisted primarily of opinions based upon varying degrees of expertise as to the cause of the leakage. In view of all the circumstances, we cannot say that the exclusion of the evidence of the crimp depth micrometer tests was harmless error, and consequently we must reverse and remand for a new trial.

We have considered the other issues raised by the appellant Crown, but find no merit in them.

That part of the judgment of the District Court fi’om which this appeal was perfected is vacated and the cause is remanded for a new trial consistent with the foregoing.

Notes

. The statutory warranties imposed by the Uniform Sales Act, in effect at the time of the transactions in controversy, provided as follows:

“(1) Where the buyer, expressly or by implication, makes known to the seller the particular purpose for which the goods are required, and it appears that the buyer relies on the seller’s skill or judgment (whether he be the grower or manufacturer or not), there is an implied warranty that the goods shall be fit for such purpose.

“(2) Where the goods are bought by description from a seller who deals in goods of that description (whether he be the grower or manufacturer or not), there is an implied warranty that the goods shall be of merchantable quality.” Tenn.Code Ann. § 47-1215 (1955 ed.).

. “Q. What conclusions were drawn by Mr. Chissom and yourself?

A. Mr. Chissom informed me that he found a number of difficulties.

He found difficulties with the soldering on the side seam.

He found infrequent but some difficulty at the crossover.

Q. What is the crossover, again?

A. That is the double juncture of the curl where you have two layers of metal here.

He also encountered some cracked beads, or either cracked or it was very difficult to pull the valve out of the can once you have crimped it, and not distort the bead, but they were elliptical, or cracked.”

. “A. At any rate I will say on Mr. Chissom’s second trip in we went through this merchandise again and it was at this point that samples of good and bad cans, and concentrate, were all sent to Philadelphia.

Mr. Chissom assured me as soon as he heard their findings he would notify me.

Shortly thereafter, a few weeks, if memory serves me correct, I received a phone call from Mr. Chissom stating that he had just heard from Philadelphia from their laboratory there, and one of the findings had to do with a deficiency of the compounding of the valve cup and that quite possibly we should bring Precision Valve Corporation, who manufactures this valve, in on this claim. I thanked him and hung up.

*925 I had just gone in to relate to Mr. Morton what I had heard when I got a subsequent phone call from Mr. Chissom saying to please disregard what he had said, that he was not supposed to have divulged this information to me, that Crown Cork & Seal made some very large percentage, in excess of eighty percent, of the valve cups that Crown used, and ultimately it would come back to Crown and for me to go ahead and file the claim with Crown.”