The issue in this case concerns the proper value for duty purposes of a model of a telephonic “small call distributing system” that was manufactured and sold by the Northern Electric Co., Ltd. of Ontario, Canada (Northern Electric), and imported into this country in June 1962 by appellee as customs broker for the account of the purchaser, Bell Telephone Laboratories of New York (Bell).
The imported model was appraised by the government on the basis of “cost of production” under section 402a(f) of the Tariff Act of 1930, as amended by the Customs Simplification Act of 1956 (19 U.S.C. § 1402(f)),1 and valued at *1390$237,482.48 (Canadian currency). The parties are in agreement that cost of production is the proper basis of appraisement but differ as to the amount of such cost of production. The dispute more specifically is whether the government appraiser properly included as part of the cost of production of the imported model design and development costs of $173,895 (Canadian currency) that were incurred prior to its exportation.
Before the trial court, the appellee-importer contended first, that no part of the design and development cost of $173,895 was part of the cost of production; and second, that in any event this design and development cost should be equally prorated between two models that the manufacturer, Northern Electric, produced, which models consisted of the imported model here in question and another model that remained in Northern Electric’s laboratory in Canada.
The trial judge rejected the appelleeimporter’s first contention that the design and development cost was not dutiable. However, he upheld its alternative contention that this design and development cost should be prorated over the two models, and thus concluded that the statutory dutiable value for the imported model was $130,576.47 (Canadian currency) rather than the appraised value of $237,482.48 (Canadian currency). Border Brokerage Co., Inc. v. United States, 65 Cust.Ct., R.D. 11725 (1970). The basis for this latter holding was expressed as follows by the trial judge:
* * * [T]he evidence establishes to my satisfaction that not only was the intent of the parties to make two models but, in actuality, two models were made. The mere fact that only one model was exported does not require the entire cost to be attributed to the model imported. A true cost of production would require the cost be prorated over the two models and I so hold.
On this appeal, appellant (defendant below) stresses that the appellee failed to establish the existence of a second model at a time preceding the exportation of the imported model, and for that reason insists that the trial judge committed error in allocating the design and development cost between those two models. The appellee (which did not file a cross-appeal) argues to the contrary that the trial judge was correct in holding that the design and development cost should be prorated between the two models. In this context, the single issue before us is whether the design and development cost is properly allocable between the two such models.
The evidence in the case is entirely documentary. It consists (among other things) of an affidavit2 of an official of Northern Electric’s research and development laboratories who stated that he was in charge of Northern Electric’s work in developing a small call distributing system under a contract between his company and Bell. The purpose of the work was to develop a design of a distributing system and to embody this design in drawings and specifications suitable to be given to a manufacturer for production purposes. Hence, the end physical product of Northern Electric’s work for Bell consisted of drawings and specifications.
The affiant stated that a technique that is used in arriving at a satisfactory *1391design of a switching system is to construct models of the system during the course of the design work and then to observe and test the models. Such models, he observed, are temporary items which are made for the purpose of insuring that the end product of the design work, i. e., the drawings and specifications, is satisfactory.
The affiant further pointed out that Northern Electric made two such models in the course of its design work for the small call distributing system. One model (the dutiable value of which is the subject of this appeal) was exported to the United States in June 1962 for the purpose of field testing on the premises of a business in Portland, Oregon. The other model remained in Northern Electric’s laboratory for further testing and design work based on the experience with the model being tested in Portland. The affiant also seated that “[bjefore production of the model which was shipped to the United States, it was contemplated that at least two models would be produced, and two were in fact produced.” He added that these two models were the only ones of the same class or kind ever manufactured by Northern Electric and that there were no other Canadian manufacturers of articles of the same kind. Lastly, the affiant declared that one set of final drawings and specifications, pursuant to the contract, was delivered to Bell and another set was retained for use by Northern Electric.
The next witness, the manager of accounting at the research and development laboratories of Northern Electric in Canada, similarly averred in an affidavit that “[bjefore production of the model which was shipped to the United States, it was contemplated that at least two models would be produced, and two were in fact produced.” He stated, in addition, that the total design and development cost (including the drawings and specifications) for the two models prior to the date of exportation of the model in question amounted to $173,895 (Canadian currency). This cost, he said, was over and above Northern Electric’s normal and usual design and development costs in its manufacturing operations.
Given this record, we find no error in the conclusion of the trial court that the design and development cost should be prorated between the two units produced rather than allocating all such cost to the model that was actually sold and exported. We start with the consideration that the affidavits in the record by qualified affiants in the employ of Northern Electric established that it was contemplated that at least two models would be produced, and two were in fact produced. Relevant in that situation is Oxford University Press, N. Y., Inc. v. United States, 36 CCPA 102, C.A.D. 405 (1949), where the issue was the allocation of production costs among articles imported or to be imported. The plaintiff had imported 5,000 copies of the Oxford Dictionary of Quotations which were appraised by the government on the basis of their cost of production. The plaintiff argued that certain nonrecurring costs for editorial work, proofreading, plates and the like should be allocated over a total of 65,000 copies which it estimated would be printed in the ordinary course of business. The government appraiser had charged all such costs to the 20,000 that had been produced at the time the 5,000 were exported. The court upheld the appraisement on the basis that there was nothing in the record from which it could even be implied that 65,000 copies of the dictionary were to be printed. As the court pointed out (36 CCPA at 105):
This appears to be a case of first impression, and we do not find in any of the authorities cited by the parties a guide for decision. It is superfluous, therefore, to discuss those cases. No one, as far as the record is concerned, prior to the date of exportation, definitely knew the number of *1392copies of the book which were to be printed. In the book publishing business, we think it is fair to assume that, as in any other business, the law of supply and demand is the measure of production. There is no evidence that there was any demand for 65,000 copies, and no basis upon which a court could determine the cost of production, except on the 20,000 copies that had been printed. It might very well have happened that the demand for the publication would not exceed 20,000 copies, and in such event the manufacturers and exporters would, as happens in many other manufacturing and selling industries, be left “holding the bag.” Such things constantly occur in the business of manufacturing, buying and selling.3 [Emphasis added.]
The clear thrust of Oxford is that development costs can be properly allocated if there is evidence showing the precise number of articles contemplated at the time of exportation. Indeed, the opinion in Oxford makes it plain that had the record in that case established that at the time of exportation the manufacturer-exporter knew that a precise number of copies of the dictionary were to be printed, the nonrecurring costs would have been prorated among that number of copies. In the present case, it is clear that before production of the model in dispute, it was intended and known that two models would be produced, and two were in fact produced. Thus, unlike the evidence in Oxford, there is no need to guess or speculate as to the number of units which were intended to be produced and which were in fact produced with the use of the design and development cost in question. Cf. F. W. Myers & Co., Inc. v. United States, 66 Cust.Ct., R.D. 11736 (1971).
It is quite true, as appellant points out, that appellee has not shown that the second model was produced prior to the exportation of the model here in issue. But there is no rule that emerges from Oxford that allocation can be only among the number of units that are produced prior to exportation; the rule that emerges rather (as we have seen) is that allocation should be made among the number which were definitely to be produced. Clearly, it is impermissible to prorate design and development costs among an indefinite number of units when it is not known how many are contemplated or intended to be produced. That kind of prorating is proscribed by Oxford. By contrast, when — as here— the number of units contemplated and intended to be produced is known prior to the manufacture of any units, proration of the design and development costs among the known number is, in our view, not only required by Oxford but necessitated by logic. For when design and development costs are incurred in producing a particular article of commerce, it makes far more sense to charge the design and development costs against the total number of units known or reasonably anticipated to be produced rather than to load the first unit with all the costs. Otherwise, the first unit would have a cost and price which would be both unrealistic and contrary to the facts of business life. To illustrate the problem more graphically, it is assumed that a manufacturer spends one million dollars in developing a particular machine. It is further assumed that the manufacturer contemplated when he started that he would produce 100 such machines seriatim, and actually did so. In such circumstances, it would be anomalous to load the million-dollar development cost onto the first machine that is imported merely because the re*1393maining 99 machines had not yet been produced. Were that done, not only would the first importer bear a disproportionate burden in having to pay duty on the total million-dollar development cost, the later importers of the remaining 99 machines would escape entirely the need to pay any duty on such development cost.
What this all comes down to is that the key question in determining whether an allocation should be made is not how many units had been produced prior to the date of exportation of the merchandise in dispute but the number of units which it was definitely known before exportation would be produced. And here, since the record establishes that even before production of the model in dispute it was intended and known that two models would be produced, and two were in fact produced, the design and development cost should be prorated between the two models actually produced. This result, it may be added, is analogous to the situation in John V. Carr & Son, Inc. v. United States, 52 CCPA 62, C.A.D. 860 (1965). There the Canadian manufacturer sought to load all fixed costs onto the articles produced for home consumption in Canada, thereby reducing the dutiable cost or value of the articles exported to the United States. The court rejected this contention, holding that the fixed costs were as much chargeable to the articles exported to the United States as those produced for home consumption in Canada. Thus the court achieved the objective of a fair allocation of the fixed costs. Likewise in the present case the trial court correctly achieved the same objective by spreading the design and development cost over the two models which it was known would be produced and were in fact produced.
We affirm the findings of fact and conclusions of law of the trial judge, which we incorporate by reference. Judgment will be entered accordingly.