More than one year prior to December 24, 1952, the date he applied for a patent on a glass reinforced thermoplastic molding compound, the inventor sold small quantities to four different purchasers. When the sales were consummated, the invention had already been reduced to practice but the product had not yet been tested in an injection molding machine; each of the four customers indicated that its purchase was for experimental purposes. At the time of the four sales, an essential ingredient— glass roving — was in short supply; for that reason, the inventor was not in a position to deliver commercial quantities of the product. The questions raised by this appeal are whether the product was “on sale” prior to December 24, 1951, within the meaning of 35 U.S.C. § 102(b),1 and if so, whether the experimental motivation for the sales nevertheless saves the patent from the statutory on sale bar.
The patent in suit issued on March 17, 1959, pursuant to an application filed on December 24, 1952.2 Product claims 1, 8 and 9 — which defendant’s compound allegedly infringes — cover “an injection molding granule containing about 15 to 60 percent by weight of glass filaments, which extend generally parallel to each other along the longitudinal axis of the granule and which are coated with a thermoplastic molding composition. The glass orientation enables the provision of a molding compound which is readily feedable in any desired quantity into the hopper of a conventional molding machine and which is injection moldable into useful glass-reinforced articles having enhanced physical properties.” Finding 13, 348 F.Supp. 1338, 1343.
The earliest corroborated date of conception of the invention was about June, 1950, when the inventor, a consulting chemist named Bradt, was first interviewed for a job with the Armorite Corporation.3 At about that time others began to consider the injection molding of glass reinforced thermoplastics. However, such efforts were confined to the use of resin-impregnated glass mat as a reinforcing material, rather than the glass roving contemplated by Bradt. This “diced mat” was unsuccessful largely because of problems associated with feeding it into an injection molding machine. Bradt’s product was superior because, unlike diced mat, it could readily be fed into the molding *1362machine. It also made it possible to injection-mold larger, stronger, and more uniform molded articles. It promptly and permanently supplanted diced mat.4
Bradt satisfied himself that his product could be readily fed into a machine by conducting a series of tests in which he forced the mixture through an orifice of the size used in injection molding machines and molds. Since the glass did not clog the orifice and the mixture passed through without separating, Bradt concluded that the product was satisfactory for injection molding.5 On the basis of these tests, Bradt claimed that there had been a sufficient reduction to practice in December, 1950, to avoid a reference dated April 25, 1951, which had been cited by the Patent Examiner.6 Bradt pointed out, however, that the material had not actually been tested in an injection molding machine because he possessed no such machine and that the commercial acceptance of *1363the product could not be known until such field tests were made.
At that time, glass roving, an essential ingredient of the product, was in short supply. For that reason, together with his lack of adequate equipment for large scale production, Bradt was not in position to deliver sufficient quantities of his compound to fill commercial orders. He did, however, make at least four separate sales before receiving his first substantial order.
In March, 1951, Victory Plastics placed an order for 500 pounds of Bradt’s roving granules at a price of 68 cents per pound. Victory Plastics was engaged in the development of a nonmetallic land mine pursuant to a military contract and purchased Bradt’s product for use in connection with that development program. In October, 1951, after encountering delay in obtaining glass roving, Bradt filled the Victory order. No conditions or restrictions of any kind were placed on Victory’s use of the product. However, as the district court found, considering the nature of the Victory Plastics’ land mine project, its 500 pound purchase was clearly an experimental or laboratory quantity.
In October, Bradt delivered “an experimental quantity of 15 pounds” of his granules to Service Plastics, Inc., a Chicago injection molder. Finding 88. Also in October, Bradt made a 20 pound shipment to P. R. Mallory; the district court found that this material was to be used only for experimentation. Finding 90. At about the same time, Bradt agreed to supply Dow Chemical Company with 200 pounds which were delivered early in 1952. Finding 87.
Bradt also supplied Hoppers Company with granules pursuant to an order for 250 pounds. Referring to this order, a letter to Bradt from a representative of the Sales Department Section of Hoppers stated that the “material is to be used for experimental molding in our laboratory and we would appreciate having it described in detail.” PX 190A.
Bradt offered the product to other potential customers at least as early as May of 1951, but the district court did not find any other consummated sales prior to the critical date of December 24, 1951.
In explanation of one of the representations to potential customers which the district court found to be “nothing more than puffing and overstatements” (Finding 86), Bradt testified:
“Q. Why did you say that then to this Cordo Chemical Company? A. Because we had an image to create and we were expecting- — we were trying to get — wait a minute, we had better go back and try to explain.
“We had this mission of getting this company Armorite on its feet and had six months to do it. We had to get materials and ideas and customers together at the same time. We had a sales problem, as well as a development problem and a manufacturing problem. This was part of the sales problem; that we knew how to physically do a job. We had to get the customer and the tests, physical tests, field experience on samples before we could get the customer for sale. So we had to bring all of these to focus at one time. This was part of our sales personality.
“We were doing a selling job * * - Our philosophy was * * * that if we had the customer and had the product developed, we could put together the equipment in the time lag that would result from between when we placed the order for the glass and the time we got the glass and get the test made.” A.243a.
There is no evidence that any purchaser, any customer who placed an order, or any potential customer whose patronage was solicited prior to December 24, 1951, was told that the purchase of the product was qualified by any restriction on use or any requirement of secrecy-
The file wrapper indicates that the commercial acceptability of the product *1364could not be known until after substantial quantities had been used in an actual injection molding machine. The district court found that such testing of the product was necessary “before a conclusion could be reached regarding its operability or feasibility,” and therefore such testing was necessary before the “invention could be completed.” Finding 78. The record does not reveal the date when the product was first used in an injection molding machine. Favorable comments were received from Victory Plastics in January, 1952, and from Koppers in March, 1952, confirming Bradt’s earlier judgment that the product would, indeed, be a commercial success.7 8Not until 1955, however, did Victory Plastics place its first order for a commercial quantity of one million pounds.
Summarizing the 1951 sales, orders and offers, the district court found:
“The transactions and offers to supply the injection molding compound claimed in the Bradt patent, which took place before January, 1952 when Victory Plastics first tested the compound in an injection molding machine, were not made for any commercial purpose but in furtherance of Bradt’s efforts to establish the workability of his concept and thereby to complete his invention. Such activities were within the ambit of experimentation and development which is permitted under 35 U.S.C. § 102(b).” Finding 92.
The court concluded that the patent was valid and that claims 1, 8 and 9 were infringed by defendant.
There are four components of plaintiff’s explanation of why § 102(b) does not invalidate the Bradt patent. First, because of the shortage of glass roving and his lack of adequate equipment in 1951, Bradt was not capable of selling the product in commercial quantities. Since the product was not “on hand,” he argues in effect that it was not “on sale.” Second, until the product had been tested in an actual machine, it was not completed and only a completed product may be placed “on sale.” Third, since the purpose of the sales was to give Bradt the benefit of tests of the product, his experimental motivation avoids the on sale bar. And finally, each of his 1951 customers was motivated by an experimental purpose, and their intent may determine whether the statute is applicable. Whether these points are considered separately or in the aggregate, we think plaintiff’s argument is predicated on an improper construction of the statute.
We first note that § 102(b) contains several distinct bars to patentability, each of which relates to activity or disclosure more than one year prior to the date of the application. Two of these — the “public use” and the “on sale” objections — are sometimes considered together although it is quite clear that either may apply when the other does not."' If either objection is raised, the burden of proving that the patented product was in “public use” or “on sale” before the critical date is on the objector. However, once such activity has been established, the burden of avoiding the statutory bar by proving a purely experimental purpose rests upon the inventor.9 The experimental use doctrine was originally developed and has been almost exclusively applied in cases *1365involving the “public use” rather than the “on sale” bar. In this case we are primarily concerned with the question whether Bradt’s product was “on sale.”
We find no merit in the argument that the product was not on sale because Bradt did not have a sufficient inventory on hand to enable him to make immediate delivery in commercial quantities. The cases which plaintiff cites to support this argument dealt with the question whether selling activity short of a completed sale was sufficient to invoke the bar.10 These cases are not followed in this circuit, see Wende v. Horine, 225 F. 501, 505 (7th Cir. 1915). The record before us discloses completed sales before the critical date as well as other selling activity. In our opinion, the patented product need not be on hand in commercial quantities for it to be “on sale” within the meaning of § 102(b).
Plaintiff’s argument that the product could not have been “on sale” because it had not yet been completed must also be rejected. In cases in which a sale has been made before the critical date, a typical issue is whether the product sold is sufficiently similar to the product described in the claims to invoke the statutory bar. See, e. g., Frantz Mfg. Co. v. Phenix Mfg. Co., 457 F.2d 314, 318-323 (7th Cir. 1972). Exact identity is not required as long as the invention is essentially completed at the time of the invalidating sale.11
In this case, the product which was sold underwent no significant change after the disqualifying sales occurred. It had already been sufficiently reduced to practice to avoid the reference dated April 25, 1951; and it subsequently achieved commercial success in the form in which it was sold. More significantly, even though Bradt’s laboratory tests could not guarantee ultimate commercial acceptance, they were sufficient to demonstrate the workability of his conception. The prior art diced mat clogged the injection molding machine because it would not satisfactorily pass through the orifice. Bradt’s tests demonstrated that his granules would not clog the machine.
We think the record rather plainly indicates that his sales were, at least in part, intended to convince the industry of the workability of his concept. The district court’s finding on the issue of completeness is ambiguous because it is not clear whether the court felt that further testing was necessary to enable the industry to reach a conclusion regarding the operability or feasibility of the product, or whether the court felt that the inventor himself was unable to reach any such conclusion.12 *1366Under the former interpretation, the court erred as a matter of law; under the latter, we reject the finding as clearly erroneous. The record plainly establishes that the completed product was on sale before the critical date.
Even though the product was completed and on sale prior to December 24, 1951, arguably the statutory bar should not apply because of the parties’ experimental motivation. For it is well settled that the “public use” bar may be avoided by proof that the use was purely experimental.13 Although plaintiff has cited no cases holding that an experimental purpose avoided the on sale bar, unquestionably such an avoidance would be appropriate in some situations. In this case plaintiff relies on the experimental purpose of (a) Bradt and (b) his customers.
The district court found that the sales and offers by Bradt in 1951 “were not made for any commercial purpose.” Our study of the record leaves us with the firm conviction that this finding is clearly erroneous. Unquestionably Bradt wanted to have the compound used in an actual injection molding machine and no such machine was available to him. But it is equally clear that his promotional activity was not confined to the objective of obtaining a test of his product. His own testimony, quoted above, clearly exposes the mixed character of his motivation. He wanted as many potential customers as possible to experiment with his product in order to develop a market for it, as well as to find out how well it would perform under commercial production. Even the public use bar would not be avoided by this type of commercial experimentation.
The experimental purposes which motivated Victory Plastics and Bradt’s other customers did not qualify the unrestricted character of his sales to them. They wanted to ascertain whether the product would be suitable for their purposes. But the terms of sale did not require them to make any report to Bradt and their work led to no changes in Bradt’s conception or in the composition which had previously been reduced to practice. Cf. Tool Research and Engineering Corp. v. Honcor Corp., 367 F.2d 449, 453 (9th Cir. 1966). Again, we recognize that a transferee’s experimentation may avoid the public us bar, particularly if there is a fiduciary relationship or contract term which entitles the inventor to the benefit of the transferee’s experimental work.14 But if the inventor has made *1367several unqualified sales of the identical product he later seeks to patent, we do not believe his customers’ experimental motivation can foreclose the application of the on sale bar.
Even though the quantities involved were small, and even though neither Bradt nor his customers could foresee the ultimate commercial acceptability of the product, we hold that the unrestricted sales to four different purchasers after the invention had been reduced to practice sufficiently to avoid an important reference cited by the Examiner and more than one year before the patent application was filed, precluded the allowance of the claims in suit.
The judgment is
Reversed.