CO WEN, Senior Circuit Judge.
Textron, Inc. (Textron) appeals from an International Trade Commission (ITC or Commission) determination that no violation of section 337 of the Tariff Act of 1930 took place in the importation from Taiwan of vertical' milling machines, parts, accessories, and attachments thereto by certain firms. Specifically, Textron claims that the ITC erred in determining that it had no common law trademark in the external ap[1021]*1021pearance of the Bridgeport Series I vertical milling machine or any portion thereof, that its asserted mark was not infringed, and that the domestic industry had not been substantially injured by those acts found to be unfair. Por the reasons set forth, we affirm.
I.
The subject matter of this case — the Bridgeport Series I vertical milling machine — is pictured herein, with its component parts labeled.
[1022]*1022It is a metal cutting machine manufactured by Textron, used to produce machined surfaces on metal by means of rotary milling cutters. It weighs approximately one ton and sells for about $6,000. The original Bridgeport vertical milling machine (similar in appearance to the present model) was designed in 1938, with a significantly different configuration from those designs then existing on the market. A number of changes have been made in the design of the machine since that time, but its general external appearance has remained essentially the same since the 1950’s.
The Bridgeport machine achieved remarkable success almost immediately after its introduction, and has held a dominant position in the vertical milling machine market for many years. The Bridgeport Series I has long been generally regarded as the standard for its size and weight in the machine tool industry — so much so that vertical milling machines of its size are often referred to as “Bridgeport-type” machines. The word “Bridgeport” has long been a federally registered trademark, and it appears prominently in bold script on the head and ram of the machine.
A small number of Bridgeport-type milling machines were imported into the United States from Taiwan as early as 1975. Many of the Taiwanese imports had a strikingly similar appearance to the Bridgeport Series I, and several had similar names, such as Kingport, Millport, Newport, and Hartford. In general, the Taiwanese machines had their names marked prominently thereon, in addition to the statement that they were made in Taiwan.
In the late 1970’s, imports of the Taiwanese machines increased sharply. The Bridgeport Series I experienced no significant decline in sales, however, until 1982, when both its domestic sales and its market share declined sharply. In October 1982, Textron filed a complaint with the ITC, alleging that 43 firms committed unfair acts and substantially injured the domestic industry by importing into the United States the look-alike Taiwanese machines, thereby violating section 337. Textron’s principal allegation was that it had a common law trademark in the overall configuration of the machine, which was infringed by a number of firms. This allegation marked the first time Textron had claimed that the Series I configuration constituted a trademark. Textron also alleged that certain firms committed other unfair acts such as infringement of its federally registered word marks “Bridgeport” and “Quill Master,” false advertising, passing off, and trademark dilution.
After a hearing in which Textron and representatives of the 34 remaining respondent firms1 presented evidence, an ITC administrative law judge (AU) made an initial determination that 20 of the firms had violated section 337. In particular, the AU found that:
1. Eleven respondents had engaged in false advertising by using photographs of the Bridgeport Series I in their sales literature;
2. Sixteen respondents had engaged in passing off their machines as Bridge-ports;
3. Textron’s registered trademark “Quill Master” had been infringed by one ' respondent in connection with an attachment to vertical milling machines.
4. Textron’s registered trademark “Bridgeport” had been infringed by one respondent which manufactured a machine it called “Bigport”;
5. One respondent falsely advertised that its vertical milling machine had patent protection; and
[1023]*10236. Respondents’ unfair acts had the effect or tendency to substantially injure the domestic industry.
Additionally, the ALJ found that the remaining 14 respondents had not violated section 337, because Textron had not proven common law trademark infringement, trademark dilution, or misappropriation of shape, design and trade dress.
Upon review, the ITC unanimously determined that none of the respondents had violated section 337. Although the Commission disagreed with a number of subsidiary findings of the AU (some of which will be discussed below), it affirmed the initial determination of the ALJ that no common law trademark infringement took place.2 The Commission also agreed with the ALJ that certain respondents had engaged in false advertising under section 43(a) of the Lanham Act, but disagreed with her finding that 16 respondents had engaged in passing off. Additionally, the ITC found that although Textron suffered a dramatic loss of sales in 1982, it had failed to prove that those acts of the respondents found to be unfair — false advertising and infringement of the “Bridgeport” and “Quill Master” registered trademarks — caused substantial injury to the domestic industry.
Textron then appealed to this court, raising the following issues:
1. Whether a common law trademark exists in the overall appearance of the Bridgeport Series I, or any portion thereof;
2. Whether, if such a common law trademark exists, it was infringed by respondents (22 of whom have intervened in this appeal); and
3. Whether the effect or tendency of those acts deemed unfair has been to destroy or substantially injure the relevant domestic industry,
II.
The ITC’s finding that no common law trademark existed in the overall configuration of the Bridgeport Series I was premised on Textron’s failure to show that the machine’s overall appearance was inherently distinctive or had acquired secondary meaning. In its discussion relating to secondary meaning (an association in buyers’ minds between the alleged mark and a single source of the product), the Commission found that Textron’s most significant pertinent evidence — the results of a survey — should be given little weight. The survey, conducted by a well known expert in the field of survey evidence, presented photographs of three different milling machines to 281 individuals who were asked whether they could identify the maker of the machines. One photograph in the survey was of a machine made by Yeong Chin of Taiwan, which appears to be nearly identical to the Bridgeport Series I. The other two photographs in the survey were of machines made by other manufacturers with considerably different appearances, which were used as controls. The labels of each of the machines were not visible in the photographs.
Survey respondents were selected throughout the United States from employees of machine shops who were familiar with vertical milling machines, which shops either owned or planned to own such machines within the next year or so. First, respondents were asked if they could identify the machine. If they answered in the affirmative, they were asked who made the machine, and why they gave this answer. Approximately 61 percent identified the Yeong Chin look-alike as a Bridgeport. [1024]*1024Only 15 and 16 percent, respectively, identified the controls as a Bridgeport. These results compare favorably with results achieved in a number of cases which have found secondary meaning for the products in question, largely on the basis of survey evidence. See, e.g., Ideal Toy Corp. v. Plawner Toy Mfg. Corp., 685 F.2d 78, 82, 216 USPQ 102,106 (3d Cir.1982) (40 percent recognition); Monsieur Henri Wines v. Duran, 204 USPQ 601, 605 (TTAB 1979) (37 percent recognition).
The Commission made a number of criticisms of the survey, which in its view justified placing little weight on the results. It agreed with the ALJ that the use of a photograph of the Yeong Chin lookalike, rather than a picture of the Series I itself, was critical, because Textron had not clearly identified the essential features of its claimed trademark. Yet the Yeong Chin machine concededly has the same overall configuration as the Series I, which in itself constitutes the principal trademark claimed by Textron. We see no reason why a look-alike item containing the same features as the claimed trademark cannot be used as a model in a survey for secondary meaning of that claimed trademark. See Ideal Toy Corp., 685 F.2d at 82, 216 USPQ at 106 (look-alike version of “Rubik’s Cube” used in survey).
The ITC also adopted the AU’s finding that the use of two vastly different machines as controls may have biased the survey results towards the selection of the Yeong Chin model as a Bridgeport. If the survey had asked the respondents to name which of the three machines was a Bridgeport, this criticism might well have merit. Because the respondents were merely asked whether they could identify the maker of each machine, however, we do not view the use of the two different controls as biasing the survey towards any particular answer. See Monsieur Henri Wines, 204 USPQ at 605. Indeed, as Textron points out, the controls serve the useful purpose of determining the extent to which the respondents’ high recognition of the Yeong Chin look-alikes was due to Bridgeport’s name recognition as the preeminent maker of vertical milling machines. The small number of respondents who identified the two controls as Bridgeports shows that recognition of the Bridgeport name alone was not a very significant factor in the high recognition of the Yeong Chin machine as a Bridgeport.
III.
Although we agree with the ITC that Textron has not shown that its machine’s design is entitled to common law trademark protection, we think this result is more properly arrived at by first applying the doctrine of functionality. A long established tenet of common law holds that trademark protection cannot be given to those product configurations deemed legally functional, even if they would otherwise be so entitled. See In re Morton-Norwich Products, Inc., 671 F.2d 1332, 1336, 213 USPQ 9, 12 (CCPA 1982) and authorities cited therein.
The reason for the functionality limitation, as explained in the seminal Morton-Norwich case, is to protect the fundamental right to compete through imitation of a competitor’s superior product, which right can only be temyorarily denied by the patent or copyright laws. Id. Morton-Norwich also recognized, however, that the public’s right to compete by copying must be balanced against the interest of protecting the goodwill built up from designs which serve to identify the maker of particular products. For this reason, this court and our predecessor court have made clear that only de jure functional designs, as contrasted with de facto functional designs, can be exempted from trademark protection. 671 F.2d at 1337, 213 USPQ at 13; In re R.M. Smith, Inc., 734 F.2d 1482, 222 USPQ 1 (Fed.Cir.1984); see also In re Deister Concentrator Co., 289 F.2d 496, [1025]*1025129 USPQ 314 (CCPA 1961). The difference between de facto and de jure functionality was simply but cogently explained by Judge Rich, author of the Morton-Norwich opinion, as follows:
In essence, de facto functional means that the design of a product has a function, i.e., a bottle of any design holds fluid. De jure functionality, on the other hand, means that the product is in its particular shape because it works better in this shape.
R.M. Smith, 734 F.2d at 1484, 222 USPQ at 3.
The case law of this court and of its predecessor also establishes that before an overall product configuration can be recognized as a trademark, the entire design must be arbitrary or non de jure functional. Petersen Mfg. Co. v. Central Purchasing, Inc., 740 F.2d 1541, 1550, 222 USPQ 562, 569 (Fed.Cir.1984); In re Minnesota Mining and Mfg. Co., 335 F.2d 836, 142 USPQ 366 (CCPA 1964). The reason for this rule is self-evident — the right to copy better working designs would, in due course, be stripped of all meaning if overall functional designs were accorded trademark protection because they included a few arbitrary and nonfunctional features. See Petersen Mfg. Co., 740 F.2d at 1550, 222 USPQ at 569; R.M. Smith, 734 F.2d at 1484, 222 USPQ at 2-3. In cases where the holders of such designs seek trademark protection, it can be obtained only for those features that are nonfunctional. Petersen Mfg. Co., 740 F.2d at 1550, 222 USPQ at 569; Vuitton et Fils S.A. v. J. Young Enterprises, Inc., 644 F.2d 769, 774, 210 USPQ 357 (9th Cir.1981). Moreover, an applicant for trademark protection has the burden to prove that a design is nonfunctional, once a prima facie case of functionality is made by an opponent. In re Teledyne, Inc., 696 F.2d 968, 971 (Fed.Cir.1982); see also R.M. Smith, 734 F.2d at 1484, 221 USPQ at 3; Morton-Norwich, 671 F.2d at 1343, 213 USPQ at 17.
Pursuant to the criteria of de jure functionality set forth in Morton-Norwich, the AU found that three of the six principal components of the Bridgeport Series I — the head, saddle and knee — are “primarily functional.” This finding was based on the de facto advantages in weight, support, and convenience of operation of these designs, as well as the lack of evidence of the existence of alternative designs that would work as well at a similar cost. With respect to the design of a fourth component of the machine — the pedestal — the AU’s findings indicate that a prima facie case of functionality was made, since the bottle shape was found to save the expense of extra metal in the pedestal. Because the AU also found that Textron did not rebut this showing with evidence that alternative designs were available or conceivable which would perform the same function at a comparable cost, we interpret her finding to be that the design of the pedestal was functional.
Although the AU also did not specifically state whether the curved design of the column and ram were de jure functional, we interpret her opinion as finding that the design of these components did not meet the Morton-Norwich functionality criteria. This is because even though it was found that the design of these parts was de facto functional, alternative designs were found to be available which would not preclude effective competition with Bridgeport if utilized. Based on the above findings, showing that the design of only two of the Series I components was nonfunctional, the AU held that Textron had not sustained its burden to show that the overall design was nonfunctional.3
The ITC disagreed with the AU’s determination on functionality because of her [1026]*1026“focus” on the component parts of the claimed mark rather than on the machine’s overall appearance. In rejecting the AU’s holding, the Commission noted that other machines with different overall configurations are able to compete with Bridgeport. It acknowledged, however, that those machines on the market with alternative configurations were at a competitive disadvantage to the Bridgeport because of the additional amounts of metal required by their designs, which added significantly to the cost of the machines. Machines with these alternative designs were able to compete with the Bridgeport, the ITC added, by making up for their weight-related cost disadvantages in other ways.
The ITC did not disagree with the AU’s findings of fact as to the functionality of the machine’s component parts, which are supported by substantial evidence. To the contrary, to the extent that it discussed the design of the machine’s components, it merely referred to evidence corroborating the AU’s finding that the curved design of the column and ram were nonfunctional. This evidence consisted of three alternative designs of the column of a vertical milling machine and one of the ram, produced as wooden models by Textron’s expert witness during the course of the evidentiary hearing. While acknowledging that the witness did not suggest alternative designs for any other components of the machine, the Commission stated that the short time necessary to make three alternative column designs “demonstrates the comparative ease” of making alternative designs for those other components.
We thus find ourselves confronted with a record devoid of actual evidence showing that any component of the Bridgeport Series I, other than the column or ram, has been or can be designed in an alternative manner and work as well, at an equivalent cost.4 In light of the strong prima facie case of de jure functionality made for the remaining components of the machine, this means that Textron has failed to carry its burden to prove that the overall configuration of its machine was nonfunctional. The Commission’s first finding on functionality — “respondents have failed to show that the Bridgeport design is essential to competition” — reflects an erroneous allocation of the burden of proof on this issue. Its ultimate holding on functionality — that Textron has sufficiently demonstrated that an alternative overall design could be produced that could perform the same functions with no significant increase in cost— is not supported by substantial evidence.
Therefore, we hold that the overall configuration of the Bridgeport Series I machine is functional, and that the only portion of the Series I design that is nonfunctional and capable of receiving trademark protection is the curved design of the column and ram. In so holding, we do not retreat from the position that the determination of whether an overall design is functional should be based on the superiority of the design as a whole, rather than on whether each design feature is “useful” or “serves a utilitarian purpose.” Teledyne, 676 F.2d at 971, 217 USPQ at 11; Morton-Norwich, 671 F.2d at 1339, 213 USPQ at 14. We merely acknowledge, as this court did in similar circumstances in Teledyne, that the best way the Commission may have had to analyze the de jure functionality of a complex overall design was to do so from the standpoint of its de facto functional features.
We also do not hold that insignificant functional element(s) in an otherwise [1027]*1027nonfunctional overall design can render that design incapable of trademark protection. Where substantially all of an overall design otherwise eligible for trademark protection is nonfunctional, as in such well known instances as the fluting of the glass in the Coca-Cola bottle, it should not matter that the shape of an insignificant element of the design, such as the lip of the bottle, is arguably functional. See 1 J. McCarthy, Trademarks and Unfair Competition § 7:31 (2d Ed.1984). We need only hold that here, where most of the product’s overall design is functional, the nonfunctional components alone, rather than the overall design, are capable of being considered a trademark.
IV.
It remains to be determined whether the design of the column and ram of the Bridgeport Series I is inherently distinctive5 or has acquired secondary meaning, thereby entitling it to trademark protection. We think that the ITC’s rejection of Textron’s claim that the machine’s overall design is inherently distinctive is equally applicable to the design of the column and the ram. Inherently distinctive marks are said to be capable of functioning immediately upon use as a symbol of origin, because they are either arbitrary or fanciful. See McCarthy, supra, §§ 11:2. Although use of the Swedish curve design may well have been unique in connection with vertical milling machines at the time Textron incorporated it into its machine, its appearance is essentially adapted to the function it performs. Such a design cannot be considered arbitrary or fanciful, devised for the sole purpose of functioning as a trademark.
We also do not think that Textron established that the design of the machine’s column and ram acquired any secondary meaning. To the extent that its principal evidence of secondary meaning (the survey) deserves to be given weight, it goes to the secondary meaning of the machine’s entire configuration, since pictures of entire vertical milling machines were shown to respondents.
Textron has not shown that a substantial number of survey respondents identified the look-alike machine as a Bridgeport because of the design of the column or ram. To the contrary, even though the 281 survey respondents were asked why they thought particular machines in the survey were made by the makers they designated, the record shows that, at most, three of them (approximately 1 percent) stated that the curved design of the column and/or ram, or even the curves of the design in general, marked any machine as a Bridgeport.6 This fact deserves particular significance when it is noted that many respondents identified the Taiwanese look-alike as a Bridgeport because of the design or position of various other components of the machine, such as the head (8.3 percent), the motor, the saddle, and the knee.
Additionally, we find Textron’s circumstantial evidence of secondary meaning for the design of the column and ram to be unpersuasive. Although Textron has pictured the silhouette of the Bridgeport machine in its operator’s manuals and certain promotional activities, it has not proffered evidence showing that these promotions focused buyers’ attentions on the shape of the machine or that the design of the column and ram was featured in any way. Cf In re Data Packaging Corp., 453 F.2d 1300, 1304, 172 USPQ 396, 399 (CCPA [1028]*10281972). Moreover, although Textron used a curved design of its column and ram for many years, we find it significant that Tex-tron did not assert that any or all of its Bridgeport design constituted a trademark until the instigation of this litigation. Given this lack of any asserted trademark in the design and the limited number of design alternatives actually in existence, we agree with the ITC that the evidence of close copying introduced by Textron is entitled to little weight.
Thus, we conclude that Textron has not demonstrated its entitlement to a common law trademark for all or any part of the design of its Bridgeport Series I machine. Because no trademark exists in this design, we need not reach the issue of whether Textron’s alleged trademark has been infringed, based on the likelihood of confusion as to source or sponsorship between the Bridgeport and the Taiwanese machines.
V.
The sole remaining issue under Textron’s section 337 claim is whether the unfair acts of the original respondents cause or have a tendency to cause substantia] injury to the domestic industry. We agree with the ITC that the unfair acts of the respondents consist solely of false advertising by 12 firms and infringement of the registered trademarks “Bridgeport” and “Quill Master” by two firms. We also must uphold the Commission’s finding that Textron did not show how these unfair acts caused or tended to cause substantial injury. The record shows that no attachments bearing the infringed “Quill Master” trademark have been imported, that a minuscule number of machines bearing the “Bigport” mark were imported in 1981-82, and that an indeterminate number of machines were sold in connection with the brochures containing deceptive advertising. Indeed, Tex-tron does not challenge the Commission’s findings that these practices did not result in a substantial loss of sales.
Textron’s only point of disagreement with the ITC’s findings of no injury from these practices is based on our decision in Bally/Midway Mfg. Co. v. United States International Trade Commission, 714 F.2d 1117, 219 USPQ 97 (Fed.Cir.1983). In that case, we held that where the unfair practice is the infringement of a domestic industry’s copyright, trademark, or patent right, even a relatively small loss of sales may establish the requisite injury to the domestic industry under section 337(a). 714 F.2d at 1124, 219 USPQ at 102. In contrast to the evidence in this case, however, the court there found that uncontra-dicted evidence established injury to Bally's Rally-X business from the unfair acts of the foreign respondents. At oral argument, counsel for Textron insisted that Bally/Midway obligates the ITC to find injury whenever a trademark has been infringed, regardless of the amount (if any) of damage caused to the domestic industry thereby. We cannot sustain such an interpretation of section 337, which would read the injury requirement entirely out of the statute whenever a trademark is found to be infringed.
Contrary to Textron’s assertion, section 337 does not function merely as the international extension of our patent, trademark and copyright laws. See In-the-Ear Hearing Aids, Tariff Commission Publ. No. 182 at 28 (July 1966). Instead, section 337 has consistently been interpreted to contain a distinct injury requirement of independent procf. Id. Congress may well have included this separate requirement in the original 1930 version of section 337 to insure that the extreme and internationally provocative remedy contemplated therein — exclusion of imports from particular countries — would be implemented only when this is compelled by strong economic reasons. See, e.g., S.Rep. No. 93-1298, 93d Cong., 2d Sess. 199 (1974); reprinted in 1974 U.S.Code Cong. [1029]*1029& Ad.News 7186, 7331, Although the contemplated range of remedies was expanded by the Trade Act of 1974 to include “softer” sanctions such as cease-and-desist orders, Congress has never altered the statute’s injury requirement. In fact, Congress expressly rejected a Nixon Administration attempt to eliminate the injury requirement in its proposed Trade Reform Act of 1973. See H. Kaye, et al., International Trade Practice § 6.05 n. 1 (1984).
We are aware of no precise and all-inclusive definition of “injury” under section 337, despite the passage of more than half a century since the enactment of the statute. This lack of precision stems in large part from the diversity of practices covered by the statute. Both this court and the ITC have acknowledged that the quantum of proof of injury is less in the context of patent, trademark, or copyright infringement, however, than in other types of unfair trade practices, because the holder of the former type of rights is entitled to exclude competitors entirely from using the intellectual property covered by those rights. See Bally/Midway, 714 F.2d at 1124, 219 USPQ at 102; In re Spring Assemblies and Components Thereof, ITC Pub. No. 337-TA-88, 216 USPQ 225, 243 (1981). Even in the context of patent, trademark or copyright infringement, the domestic industry must normally establish that the infringer holds, or threatens to hold, a significant share of the domestic market in the covered articles or has made a significant amount of sales of the articles. In re Spring Assemblies, 216 USPQ at 243.
Textron has not shown that sales of the articles which infringe its registered trademarks are sufficient to cause, or threaten to cause, injury to the domestic market, or that the articles occupy, or threaten to occupy a significant share of the market. Moreover, under the stricter injury standard applied for other unfair practices, such as false advertising, Textron’s proof, under the standards employed by the ITC, falls short of that required for a determination of injury.
CONCLUSION
The determination of the ITC that there has been no violation of section 337 by any of the 34 respondents is affirmed.
AFFIRMED.