This is an appeal from the district court's issuance of a preliminary injunction enjoining defendants Kohler Company and Robern, Inc. from selling the Kohler Falling Water faucet, a faucet resembling plaintiff Lund's VOLA faucet. The VOLA faucet mounts on a wall, has been in the design collection at the Museum of Modern Art, and has a certain cachet among those who enjoy bathrooms and kitchens beautiful. Kohler intended to produce a faucet like Lund's, but not identical to it, and hence designed the now-enjoined Falling Water faucet. There were two basic claims before the trial court: that the Falling Water faucet "diluted" the VOLA faucet's trade dress within the meaning of the Federal Trademark Dilution Act of 1995 ("FTDA"), 15 U.S.C. § 1125(c), and that the Falling Water faucet infringed the VOLA's trade dress. Lund won a preliminary injunction on the first ground, but not on the second. See I.P. Lund Trading ApS v. Kohler Co., 11 F. Supp. 2d 112, 127 (D.Mass.1998) ("Lund I ").

This difference in results was not anomalous. The district court found, as to the infringement claim, that while the VOLA faucet was not inherently distinctive, it had acquired secondary meaning and thus was protectable, but that there was no infringement because there was no confusion on the part of consumers. In contrast, under the FTDA, where no confusion need be shown, the court found the VOLA faucet was famous and that Kohler's faucet diluted the identity of the VOLA faucet.

This case presents complex issues arising in areas of intellectual property law recently extended and not yet well demarcated. Few courts of appeals have yet interpreted the FTDA and this court has never addressed certain key issues, under both the infringement and FTDA claims, necessary to the resolution of the case. The district court wrote thoughtfully, and, particularly as to the FTDA issues, without much appellate guidance. The claim for protection here comes not from traditional marks such as names but from the very design of the faucet itself--that design is said to give the faucet its identity and distinctiveness. Although Lund may have been able to obtain a design patent and so protect its VOLA faucet in that way, at least for a period of fourteen years, see 35 U.S.C. § 173, it chose not to. Rather, it chose to turn for protection to legal doctrines of trademark and trade dress, originally crafted without product designs in mind. The trade dress of product designs, unlike other forms of trade dress, cannot be separated from the product itself. Kohler has raised serious constitutional concerns, saying that this use of the FTDA against a competing product essentially gives a perpetual monopoly to product design, a perpetual monopoly prohibited by the Patent Clause.

Kohler and Robern (collectively "Kohler") argue that the district court erred in its determination that plaintiffs I.P. Lund Trading ApS and Kroin Incorporated (collectively "Lund") demonstrated a likelihood of success on the merits of their claim under the FTDA that the Falling Water faucet dilutes the trade dress of Lund's VOLA faucet.1 Lund cross-appeals, arguing that the district court erred in determining that Lund was unlikely to succeed on the merits of its infringement claim. We affirm the denial of the preliminary injunction on the infringement claim. We vacate the grant of the injunction on the FTDA claim.

Several questions of first impression are resolved in this opinion. We hold that the burden of showing non-functionality of a product for which trade dress protection is sought rests on the party seeking that protection. Here that is the plaintiff Lund. In analyzing inherent distinctiveness in the context of product design, we hold that while the well-known Abercrombie test provides a useful analogy, strict application of the test is not required; we reiterate this court's adherence to the Seabrook Foods test. We emphasize that, in any case where the trade dress is said to arise from the product design, there must be separate analyses as to (1) whether a design is inherently distinctive and (2) whether it has nonetheless acquired distinctiveness through secondary meaning. As to secondary meaning said to stem from the design of the product itself, we hold that the plaintiff must show that the primary significance of the design is to signify its source.

Under the FTDA, we hold that a party who wishes to establish fame of the trade dress for which protection is sought bears a significantly greater burden than the burden of establishing distinctiveness for infringement purposes. The FTDA creates an exceptional anti-dilution remedy for truly famous marks. Once this greater burden of establishing fame has been met under the FTDA, the issue of dilution must be addressed. We reject the use of the "Sweet factors" as the test for dilution and instead require an inquiry into whether target customers will perceive the products as essentially the same. We hold that the dilution standard is a rigorous one, and Lund has not shown that it is likely to succeed. While we acknowledge serious constitutional concerns about application of the FTDA to a dilution claim against a competing product which does not confuse consumers, the resolution of the case obviates the as applied constitutional issue, and we decline to address any residual facial challenge.

The district court "enjoys considerable discretion" in determining whether to grant a preliminary injunction, but its decision "must be supported by adequate findings of fact and conclusions of law." TEC Eng'g Corp. v. Budget Molders Supply, Inc., 82 F.3d 542, 544-45 (1st Cir.1996); see also Camel Hair & Cashmere Institute of America, Inc. v. Associated Dry Goods Corp., 799 F.2d 6, 12-13 (1st Cir.1986). "On appellate review of the grant or denial of a preliminary injunction, the deferential standard of 'abuse of discretion' applies to judgment calls, by the district court, such as those that involve the weighing of competing considerations." Public Serv. Co. v. Patch, No. 98-1764, --- F.3d ----, 1998 WL 823177, at * 5 (1st Cir. Dec. 3, 1998). As explained in Ocean Spray Cranberries, Inc. v. Pepsico, Inc., 160 F.3d 58 (1st Cir.1998), "[t]he usual rubric refers to abuse of discretion ... but this phrasing is most pertinent to issues of judgment and the balancing of conflicting factors; rulings on abstract legal issues remain reviewable de novo, and findings of fact are assessed for clear error." Id. at 61 n. 1 (citations omitted). If findings are made under incorrect standards, little or no deference is due those findings. Cf. Uno v. City of Holyoke, 72 F.3d 973, 978 (1st Cir.1995). Further, "[a]buse of discretion occurs ... when a material factor deserving significant weight is ignored, when an improper factor is relied upon, or when all proper and no improper factors are assessed, but the court makes a serious mistake in weighing them." Foster v. Mydas Assocs., Inc., 943 F.2d 139, 143 (1st Cir.1991) (internal quotation marks omitted).

A party seeking a preliminary injunction must establish that (1) it is substantially likely to succeed on the merits of its claim; (2) absent the injunction there is "a significant risk of irreparable harm"; (3) the balance of hardships weighs in its favor; and (4) the injunction will not harm the public interest. TEC Eng'g Corp., 82 F.3d at 544 (discussing a claim for trade dress infringement). In the trademark context, "irreparable harm may be shown even in the absence of actual injury to plaintiff's business based on plaintiff's demonstration of a likelihood of success on the merits on its claim." Calamari Fisheries, Inc. v. The Village Catch, Inc., 698 F. Supp. 994, 1013 (D.Mass.1988) (citing Camel Hair & Cashmere Institute, 799 F.2d at 14). There is no argument that the district court applied the wrong test for injunctive relief; there is considerable dispute over the subsidiary tests which the court applied in determining whether there was probability of success on the various elements of the claims.

Lund, a Danish corporation, manufactures bathroom and kitchen fixtures and accessories, including faucets. Lund has been a family-owned corporation since its establishment in 1873. In 1969, Lund introduced the VOLA faucet, designed by the noted architect Arne Jacobsen. The faucet, which has received numerous awards over the past quarter-century, is Lund's principal revenue-producing product. Lund has sold a total of more than 600,000 VOLA faucets. The faucet has been regularly advertised and featured in numerous magazines. Kroin Incorporated is the sole United States distributor of the VOLA.

Kohler is the largest supplier of plumbing fixtures in this country, selling hundreds of types of kitchen and bathroom fixtures. In 1994, Kohler contacted Lund regarding the possibility of selling the VOLA faucet under Kohler's name. In 1995, Kohler purchased eight VOLA faucets from Lund for the purpose of testing the faucets to see if they fit in a sink that Kohler planned to introduce and to ensure that the faucets complied with United States regulations. Kohler claims that it tested the faucets and found that they did not meet U.S. regulations regarding water flow capacity and resistance to hydrostatic pressure, a contention Lund contests, and as to which there was conflicting evidence.

Kohler gave a VOLA faucet to Erich Slothower, an industrial designer employed by Kohler. Slothower then designed the Falling Water faucet, which Kohler introduced for sale at a price lower than that of the VOLA faucet. Slothower testified that he examined the VOLA carefully prior to designing the Falling Water, but that he attempted to make the Falling Water faucet different from the VOLA. Kohler's large size and well-established distribution channels mean that the Falling Water is likely to be more easily available, in addition to being less expensive.

The district court found a number of similarities between the VOLA and the Falling Water faucets. Both are "single-control, wall-mounted faucets" with handles that "utilize a thin cylindrical lever to adjust water temperature and volume"; both have "spouts and aerator holders ... of uniform diameter," with the spouts "bend[ing] downward at right angles softened by a curve"; and "both faucets offer spouts in almost exactly the same three lengths." Lund I, 11 F. Supp. 2d at 116. Both faucets fit no-hole sinks.2 In contrast, most sinks sold in the United States are "three-hole" sinks, with one hole each for the water spout and the hot and cold spigots. However, the district court also found dissimilarities between the faucets, including differences in the faucets' handles, a rounded lever on the Falling Water faucet compared to a flat lever on the VOLA, and a rounded bonnet--a piece that connects the faucet to the wall--on the mounting end of the Falling Water spout, compared to no bonnet on the VOLA. See id. The court also found that the housemarks, "VOLA" and "Kohler," are clearly dissimilar and are prominently displayed on the faucets. See id. at 123.

Co-defendant Robern, which Kohler acquired in August 1995, also purchased a number of VOLA faucets. Before being acquired by Kohler, Robern purchased 218 VOLA faucets from Kroin for use in a sink module. Robern apparently promoted its sink module pictured with the VOLA faucets. At approximately the same time as Kohler acquired Robern, Kroin refused to sell additional VOLA faucets to Robern, claiming that Robern was selling the faucets to Kroin's customers at prices below those Kroin was charging. One year later, Robern announced plans to market its sink module with the Falling Water faucet. Lund produced evidence that Robern has continued to use pictures of the VOLA in promotional materials, despite the fact that it has replaced the VOLA with the Falling Water faucet in its sink modules.

Kohler introduced the Falling Water faucet to the market in 1996. Lund filed suit on February 27, 1997, alleging trade dress infringement under Section 43(a) of the Lanham Act and trade dress dilution under the FTDA.3 Kohler denied any violations and argued that the FTDA was unconstitutional as applied to product designs. The district court held an evidentiary hearing on April 16 and 30, 1997.

The court ruled on the motion for a preliminary injunction in three stages. In a February 5, 1998 Memorandum and Order, the court found that Lund had demonstrated a substantial likelihood of success on its dilution claim, and so enjoined Kohler from selling its Falling Water faucet. See Lund I, 11 F. Supp. 2d at 127. At the same time, the court found that Lund had failed to show likelihood of success on its claim of trade dress infringement, and it reserved decision on the question of the constitutionality of the FTDA pending additional briefing. See id.

On February 12, 1998, the district court stayed the preliminary injunction pending resolution of Kohler's claim that application of the FTDA to product designs violates the Patent Clause.

In an order dated March 31, 1998 and a memorandum dated April 2, 1998, the district court found that Kohler's constitutional argument was unlikely to succeed on the merits, lifted the stay on the injunction, and ordered Lund to post a bond of $250,000. See I.P. Lund Trading ApS v. Kohler Co., 11 F. Supp. 2d 127, 134-35 (D.Mass.1998) ("Lund II ").

The district court also denied Kohler's request to stay the injunction pending appeal to this court. On April 29, 1998, this court rejected Kohler's request for a stay pending appeal.

III. Purposes of Trademark and Trade Dress Protection

The basic building blocks of the analysis are worth reiterating. Section 43(a) of the Lanham Act provides protection against the use of "any word, term, name, symbol, or device" that "is likely to cause confusion, or to cause mistake, or to deceive" as to the source of a product. 15 U.S.C. § 1125(a). Trade dress includes "the design and appearance of [a] product together with the elements making up the overall image that serves to identify the product presented to the consumer." Chrysler Corp. v. Silva, 118 F.3d 56, 58 (1st Cir.1997) (alteration in original) (quoting Fun-Damental Too, Ltd. v. Gemmy Indus. Corp., 111 F.3d 993, 999 (2d Cir.1997)) (internal quotation marks omitted). The new law, the FTDA, Lanham Act section 43(c), grants protection to "famous" marks against any use of the mark that "causes dilution of the distinctive quality of the mark." 15 U.S.C. § 1125(c).

To resolve the issues presented, we go back to the underlying purposes of trademark and trade dress infringement and dilution protections. The various intellectual property protection mechanisms serve related but distinct ends. These distinct ends inform the selection of appropriate tests under the different sections of the Lanham Act. These distinctions are also particularly pertinent to Kohler's constitutional claim that dilution protection of trade dress of product design amounts to an unconstitutional perpetual monopoly under the Patent Clause of the Constitution. The Patent Clause itself describes the "exclusive Right" given as being for "limited Times." U.S. Const. art. I, § 8, cl. 8. "The laws of patents, copyright, trade secrets, trademarks, unfair competition, and misappropriation balance the conflicting interests in protection and dissemination differently in different contexts through specific rules that determine just who will receive protection, of just what kind, under what circumstances, and for how long." DeCosta v. Viacom Int'l, Inc., 981 F.2d 602, 605 (1st Cir.1992).

A primary purpose of trade dress or trademark protection is to protect that which identifies a product's source. See Qualitex Co. v. Jacobson Products Co., 514 U.S. 159, 162, 115 S. Ct. 1300, 131 L. Ed. 2d 248 (1995) (noting that "the statutory definition of a trademark ... requires that a person 'us[e]' or 'inten[d] to use' the mark 'to identify and distinguish his or her goods, including a unique product, from those manufactured or sold by others and to indicate the source of the goods, even if that source is unknown' " (alterations in original) (quoting 15 U.S.C. § 1127)). "The purpose of trademark laws is to prevent the use of the same or similar marks in a way that confuses the public about the actual source of the goods or service." Star Fin. Servs., Inc. v. Aastar Mortgage Corp., 89 F.3d 5, 9 (1st Cir.1996). Traditional trademark and trade dress law thus encourages production of products of high quality "and simultaneously discourages those who hope to sell inferior products by capitalizing on a consumer's inability quickly to evaluate the quality of an item offered for sale." Qualitex, 514 U.S. at 164, 115 S. Ct. 1300. More importantly for our purposes, trademark and trade dress protection serves to protect both the trademark or trade dress owner and the public by avoiding confusion or mistake.

In contrast, dilution statutes, and the FTDA in particular, protect only the trademark or trade dress owner and are not concerned with possible confusion on the part of consumers. See J. Gilson, Trademark Protection and Practice § 5.12 (1998). Any protection of the public intended by the FTDA is indirect at best. "Anti-dilution statutes have developed to fill a void left by the failure of trademark infringement law to curb the unauthorized use of marks where there is no likelihood of confusion between the original use and the infringing use." L.L. Bean, Inc. v. Drake Publishers, Inc., 811 F.2d 26, 30 (1st Cir.1987). Filling this void, Congress passed the FTDA in 1995 both to provide uniform national protection against dilution and to bring this country's law into conformity with international agreements. See H.R.Rep. No. 104-374, at 3-4 (1995), reprinted in 1995 U.S.C.C.A.N 1029, 1030-31.4

IV. Prerequisites for Protection from Infringement and Dilution

Despite different purposes being served, claims for protection against trademark and trade dress infringement, on the one hand, and dilution, on the other, share three common elements before the analyses diverge. Those elements are that marks (a) must be used in commerce, (b) must be non-functional, and (c) must be distinctive. While all such marks may be protected against infringement, under the FTDA only famous and distinctive marks are eligible for protection against dilution. No requirement for fame is present in trademark and trade dress infringement.

There is no dispute that both Lund's and Kohler's faucets have been used in commerce, thus satisfying the first requirement for protection against both infringement and dilution. See 15 U.S.C. § 1125(a),(c).

To be protected under the Lanham Act, a trademark or trade dress must not be functional. See Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 775, 112 S. Ct. 2753, 120 L. Ed. 2d 615 (1992). If the trade dress is functional, it receives no protection under trademark law. The functionality doctrine has considerable economic and legal significance. "The functionality doctrine prevents trademark law, which seeks to promote competition by protecting a firm's reputation, from instead inhibiting legitimate competition by allowing a producer to control a useful product feature." Qualitex, 514 U.S. at 164, 115 S. Ct. 1300. Thus, the functionality doctrine marks the boundaries of trade dress protection.

As discussed later, "if functional features were given trademark or trade dress protection, such protection would clearly clash with the objectives of federal functional patent law." 1 J. McCarthy, McCarthy on Trademarks and Unfair Competition § 7:64 (4th ed.1996). The rule against functional features being protected as symbols of origin "is obviously to prevent the grant of perpetual monopoly by the issuance of a trade-mark in the situation where a patent has either expired, or ... cannot be granted." Sylvania Elec. Prods., Inc. v. Dura Elec. Lamp Co., 247 F.2d 730, 732 (3d Cir.1957).

The core inquiry into whether trade dress is functional requires examination of the effects that granting protection to a product will have on the ability of others to compete. Thus, in Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844, 102 S. Ct. 2182, 72 L. Ed. 2d 606 (1982), the Court stated that a functional product feature is one that "is essential to the use or purpose of the article or [that] ... affects the cost or quality of the article." Id. at 851 n. 10, 102 S. Ct. 2182. In Qualitex, the Court added that the inquiry into functionality turns in part on whether granting protection to a mark "would permit one competitor ... to interfere with legitimate (nontrademark-related) competition through actual or potential exclusive use of an important product ingredient." Qualitex, 514 U.S. at 170, 115 S. Ct. 1300.5

The fact that a product contains some functional elements does not, however, preclude Lanham Act protection. "[A] particular arbitrary combination of functional features, the combination of which is not itself functional, properly enjoys protection." Taco Cabana Int'l, Inc. v. Two Pesos, Inc., 932 F.2d 1113, 1119 (5th Cir.1991), aff'd, 505 U.S. 763, 112 S. Ct. 2753, 120 L. Ed. 2d 615 (1992). The crucial inquiry is into the effect that granting protection will have on the opportunity of others to compete.

As the parties note, this court has not previously decided whether a showing of non-functionality is an element of the claim of the party seeking protection, or whether functionality is an affirmative defense on which the defending party has the burden. See TEC Eng'g Corp., 82 F.3d at 546 n. 3. In Fisher Stoves, Inc. v. All Nighter Stove Works, Inc., 626 F.2d 193 (1st Cir.1980), this court, although not resolving the issue, stated that it was "not at all clear" that the district court erred when it placed the burden of proof on the plaintiff. Id. at 195. Other circuit courts which have decided the issue are split. Compare Sunbeam Prods., Inc. v. West Bend Co., 123 F.3d 246, 251 (5th Cir.1997) (holding that plaintiffs bear the burden), cert. denied, --- U.S. ----, 118 S. Ct. 1795, 140 L. Ed. 2d 936 (1998), Woodsmith Publishing Co. v. Meredith Corp., 904 F.2d 1244, 1247 (8th Cir.1990), Rachel v. Banana Republic, Inc., 831 F.2d 1503, 1506 (9th Cir.1987), AmBrit, Inc. v. Kraft, Inc., 812 F.2d 1531, 1535 (11th Cir.1986), Kwik-Site Corp. v. Clear View Mfg. Co., 758 F.2d 167, 178 (6th Cir.1985), and CIBA-GEIGY Corp. v. Bolar Pharmaceutical Co., 747 F.2d 844, 854 (3d Cir.1984), with Computer Care v. Service Sys. Enters., Inc., 982 F.2d 1063, 1068 (7th Cir.1992) (holding that defendants bear the burden), Brunswick Corp. v. Spinit Reel Co., 832 F.2d 513, 520 (10th Cir.1987), and LeSportsac, Inc. v. K Mart Corp., 754 F.2d 71, 75-76 (2d Cir.1985). See generally 1 McCarthy § 7:72. The Lanham Act itself provides no guidance on the issue, and the caselaw contains little discussion of the rationale for allocating the burden of proof to plaintiffs or defendants.

We hold that the party alleging trademark infringement and dilution bears the burden of proving non-functionality of those elements of the physical object that the plaintiff claims constitute the mark and for which the plaintiff is seeking protection. Several rationales support the burden being placed on those seeking protection, here the plaintiff. Functionality, or, more precisely, a showing of non-functionality of the elements for which protection is sought, is an essential component of the protection the law gives to trademarks and trade dress. A showing of non-functionality is essential because the doctrine prevents trademarks from limiting legitimate competition. Put differently, "functionality" plays an important role in preventing a constitutional problem between the Lanham Act and patent law. Moreover, Congress did not intend to provide Lanham Act protection to functional aspects of products, see Two Pesos, 505 U.S. at 775, 112 S. Ct. 2753, and thus it would be anomalous if the burden were not placed on the party seeking protection. Such an assignment of the burden of proof ensures that trademarks serve their intended purpose of identifying product sources.

A plaintiff's product is known best by plaintiff itself. A contrary result, placing the burden on defendant, could lead to untoward results. It could also lead to unwarranted litigation. If protected elements of a product are indeed functional, and the burden of showing non-functionality is placed on plaintiff, plaintiff will have incentives not to bring unwarranted actions hoping that defendants will fail to raise or prove functionality as an affirmative defense. It would be far easier for plaintiffs to obtain protection for functional aspects of their products--and thus limit legitimate competition--if defendants were to bear the burden of proof, as defendants might lack the ability or incentive to pursue the issue fully. There is value in minimizing the chances of the issue not being properly raised or presented.

The district court did not make a specific finding on functionality and the record on the important issue of functionality is unclear.6 Although Lund I commented that "[n]o one argues that the VOLA design is 'functional,' as that term is used in trademark law," Lund I, 11 F. Supp. 2d at 120 n. 12, Kohler did in fact argue in a filing that "to the extent that some features on the respective faucets may somewhat resemble each other, these features are either functional in nature and thus not protectable under the Lanham Act or so commonly used in the plumbing industry that they are obvious--not distinctive of a particular product" (emphasis added). Thus Kohler did not waive the issue. We cannot say that it is obvious no aspect of the VOLA is functional; indeed some aspects give the appearance of being functional and so the issue is a real one. There are significant dissimilarities between the two faucets, and at least some of the similarities that do exist are suggestive of functionality. Nonetheless, we will assume at this stage of the case that there is some non-functional residuum based on the aesthetic unity and proportions of the VOLA.

The issue of functionality plays a key role in this case. In the absence of a finding of non-functionality of the aspects of the VOLA for which protection is sought, and in light of our placing the burden of proof on plaintiff, there can be no trademark or trade dress protection. If the VOLA design or aspects of it are functional, then the only source of exclusive rights would be in a utility patent. Trademark and trade dress law cannot be used to evade the requirements of utility patents, nor the limits on monopolies imposed by the Patent Clause. This also means that in the absence of a finding of non-functionality, no protection is available under the FTDA and no injunction may issue under the FTDA, a point to which we return later.

But there is another reason that the functionality analysis must be undertaken. There is a relationship between functionality and distinctiveness:

If a feature is functional, it is likely that all similar articles will have a similar functional feature, and one seller's feature is not likely to evoke any response in buyers that it is unique or is a distinctive symbol of origin.

1 McCarthy § 7:64. It is also clear that even if a functional feature has achieved secondary meaning as an indication of origin, that feature is not protectable under trademark or trade dress law. See Fisher Stoves, 626 F.2d at 195-96; 1 McCarthy § 7:66.

The third prerequisite for protection is distinctiveness. In order to receive trade dress protection, a product must either be inherently distinctive or have acquired secondary meaning. See Two Pesos, 505 U.S. at 769, 112 S. Ct. 2753; TEC Eng'g Corp., 82 F.3d at 545. The inquiry into distinctiveness turns on the total appearance of the product, not on individual elements. Cf. Sunbeam, 123 F.3d at 251 (stating that " 'trade dress' refers to the total image and overall appearance of a product"). The district court found, within the usual preliminary injunction standards about likelihood of success, that the VOLA faucet was not inherently distinctive but that it had acquired secondary meaning.

In analyzing whether a product's mark is distinctive, courts have often divided marks into the five categories set forth in Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4 (2d Cir.1976): (1) generic, (2) descriptive, (3) suggestive, (4) arbitrary, and (5) fanciful. See Two Pesos, 505 U.S. at 768, 112 S. Ct. 2753 (discussing Abercrombie ); Abercrombie, 537 F.2d at 9. Suggestive, arbitrary, and fanciful marks are deemed inherently distinctive; descriptive marks receive protection only upon a showing that they have acquired secondary meaning; and generic marks are not protectable. Abercrombie itself addressed words, the usual form of mark. The Abercrombie test has been used in some other areas of trade dress, such as product packaging and the overall appearance of a restaurant. See Two Pesos, 505 U.S. at 768-69, 773, 112 S. Ct. 2753.

Courts have struggled with whether the Abercrombie test, originally designed for words, should be imported wholesale into that specialized area of trade dress claimed to come from product design. Although the Supreme Court in Two Pesos endorsed the Abercrombie test in the context of non-verbal trade dress not involving product designs, at least two circuits have been skeptical of the appropriateness of the test in the product design context. See Knitwaves, Inc. v. Lollytogs Ltd., 71 F.3d 996, 1007-08 (2d Cir.1995); Duraco Prods., Inc. v. Joy Plastic Enters., Ltd., 40 F.3d 1431, 1445-49 (3d Cir.1994); 1 McCarthy § 8:12. Indeed, these courts, questioning whether a product design can ever be inherently distinctive, apply a more rigorous standard for determining inherent distinctiveness of product designs than they do for determining the inherent distinctiveness of more traditional forms of trade dress. See Knitwaves, 71 F.3d at 1008-09 (requiring that a product design serve primarily to indicate a product's source); Duraco Prods., 40 F.3d at 1449 (requiring product designs to be "(i) unusual and memorable; (ii) conceptually separable from the product; and (iii) likely to serve primarily as a designator of origin of the product"). The problem with applying traditional trademark or trade dress classifications to product designs arises because "one cannot automatically conclude from a product feature or configuration--as one can from a product's arbitrary name ...--that, to a consumer, it functions primarily to denote the product's source." Duraco Prods., 40 F.3d at 1441.

Parting from the skeptics, the Eighth Circuit has rejected any differentiation between product designs and other forms of trade dress. See Stuart Hall Co. v. Ampad Corp., 51 F.3d 780, 787-88 (8th Cir.1995). The Stuart Hall court argued that the Second and Third Circuit tests would effectively eliminate the possibility that a product design could be inherently distinctive, collapsing the inherent distinctiveness inquiry into the secondary meaning inquiry. See id. at 787 ("The requirement of source-identification applies not to whether a trade dress is inherently distinctive, but to whether it has a secondary meaning."). Such an outcome would directly contradict the Supreme Court's holding in Two Pesos that inherently distinctive trade dress is protectable even absent a showing of secondary meaning. See id. at 788 (citing Two Pesos, 505 U.S. at 769-70, 112 S. Ct. 2753).

The district court here applied the Second Circuit's Knitwaves test. It examined whether the VOLA's design was "likely to serve primarily as a designator of origin of the product." Lund I, 11 F. Supp. 2d at 120 (quoting Knitwaves, 71 F.3d at 1008) (internal quotation marks omitted). Like the court in Knitwaves, the district court did so by looking to whether the VOLA's creators intended the design to serve as a source indicator. See id.; see also Knitwaves, 71 F.3d at 1009 ("As Knitwaves' objective in the two sweater designs was primarily aesthetic, the designs were not primarily intended as source identification. Those sweater designs therefore fail to qualify for protection of trade dress inherent in product design."). The court found that the VOLA's design was "not 'primarily' intended as source identification" because "Lund's design objective was 'primarily aesthetic,' " and thus "the design of the VOLA faucet cannot be considered inherently distinctive." Id.

Lund argues that the district court was in error because it was obligated by Supreme Court precedent to apply the Abercrombie test. Lund argues that application of the wrong test produced the wrong result. Lund points to other decisions in the District of Massachusetts which have rejected the Knitwaves and Duraco Products tests and adopted the Abercrombie analysis in the product design context. See Big Top USA, Inc. v. Wittern Group, 998 F. Supp. 30, 46-47 (D.Mass.1998); Lainer v. Bandwagon, Inc., 983 F. Supp. 292, 300 (D.Mass.1997).

We do not believe that the Supreme Court's endorsement of the Abercrombie test in Two Pesos requires a strict application of the Abercrombie test in all contexts, particularly where product design is involved. The Supreme Court stated only that "[m]arks are often classified in" the five Abercrombie categories. Two Pesos, 505 U.S. at 768, 112 S. Ct. 2753. The Court did not mandate the application of the Abercrombie test; rather, it affirmed the use of the Abercrombie factors in the case before it. See id. at 768-69, 112 S. Ct. 2753. The holding of Two Pesos was that plaintiffs seeking protection for inherently distinctive trade dress are not required to demonstrate secondary meaning. See id. at 770, 112 S. Ct. 2753. Unless the Court decides to carve out an exception to this rule for claims about product design, which we deem unlikely, then the holding must be honored. And so we agree with the Eighth Circuit that the test for inherent distinctiveness should not be altered to the degree that it eviscerates the distinction between inherently distinctive trade dress and trade dress that has acquired secondary meaning. We do not believe, however, that analysis of the problem using different factors than the Abercrombie factors results in such an outcome.

This court has previously relied on the test set forth in Seabrook Foods, Inc. v. Bar-Well Foods Ltd., 568 F.2d 1342 (C.C.P.A.1977), to determine whether a product design is inherently distinctive, and we do so again. In Wiley v. American Greetings Corp., 762 F.2d 139 (1st Cir.1985), this court stated that inherent distinctiveness of a product design should be determined by reference to

whether [the design] was a "common" basic shape or design, whether it was unique or unusual in a particular field, whether it was a mere refinement of a commonly-adopted and well-known form of ornamentation for a particular class of goods viewed by the public as dress or ornamentation for the goods, or whether it was capable of creating a commercial impression distinct from the accompanying words.

Id. at 141 (alteration in original) (quoting Seabrook Foods, 568 F.2d at 1344) (internal quotation marks omitted). We also agree with one commentator's analysis that "[i]n reality, all three [Seabrook Foods ] questions are merely different ways to ask whether the design, shape or combination of elements is so unique, unusual or unexpected in this market that one can assume without proof that it will automatically be perceived by customers as an indicator of origin--a trademark." 1 McCarthy § 8:13. Although the plaintiff in Wiley brought suit under state common law, we find the test equally applicable to claims brought under the Lanham Act.

The Seabrook Foods test is largely consistent with the Second Circuit's Knitwaves test for inherent distinctiveness, that is, whether the design "is likely to serve primarily as a designator of origin of the product." Knitwaves, 71 F.3d at 1008 (quoting Duraco Prods., 40 F.3d at 1449) (internal quotation marks omitted); see also Landscape Forms, Inc. v. Columbia Cascade Co., 113 F.3d 373, 378 n. 3 (2d Cir.1997) (rejecting an argument that the Knitwaves and Seabrook Foods tests are inconsistent).

In contrast to the Second Circuit, however, we believe that Two Pesos obliges courts to maintain a clear distinction between the inquiry into secondary meaning and the inquiry into inherent distinctiveness. Cf. Knitwaves, 71 F.3d at 1008. Although plaintiffs seeking to demonstrate that a product design is inherently distinctive may find that their task is a difficult one, Two Pesos makes clear that this opportunity remains open. Product designs are unlike many more traditional subjects of trademark protection, "which almost automatically tell a customer that they refer to a brand," Qualitex, 514 U.S. at 162-63, 115 S. Ct. 1300, and it is therefore unlikely that many product designs will be found to be inherently distinctive. Nevertheless, plaintiffs seeking trade dress protection for product designs are entitled to attempt to show that their designs are so unique as to be primarily perceived as indicating the product's origin. Cf. 2 McCarthy § 15:9 ("A design or shape, or a trade dress that is so fanciful as to be inherently distinctive can function as a mark without the need for proof of secondary meaning. The issue is whether this shape is so different or unusual for this type of goods or services that its distinctiveness can be assumed.").

Applying Knitwaves, the district court did make a finding as to inherent distinctiveness. It found that the VOLA's design is not "likely to serve primarily as a designator of origin of the product." On this record that finding cannot be said to be clearly erroneous. One aspect of the test applied in reaching this conclusion is, nonetheless, problematic. The district court relied on evidence as to Lund's intent. But the plaintiff's intent is not entitled to much weight in a determination of whether a product design is inherently distinctive. See 1 McCarthy § 8.13 (stating that "the intent of the designer is a very weak indicator of the likely reaction of potential customers"); cf. Landscape Forms, 113 F.3d at 377 n. 3 ("If Knitwaves forced courts to decide whether a manufacturer's purpose was to create either something of beauty or something indicative of source, we agree the task would often prove impossible."). The district court may, however, have given little weight to this factor in reaching its conclusion.7

The district court used the Knitwaves test, which is consistent with this Circuit's Wiley /Seabrook test, to reach its conclusion. There is adequate evidence in the record to support the district court's determination there was little probability of success in demonstrating that the VOLA's design inherently and primarily serves to identify the faucet's source.

The second method for demonstrating that a product's trademark or trade dress is distinctive and thus protectable is through a showing that the mark has acquired secondary meaning. "To establish secondary meaning, a manufacturer must show that, in the minds of the public, the primary significance of a product feature or term is to identify the source of the product rather than the product itself." Inwood Laboratories, 456 U.S. at 851 n. 11, 102 S. Ct. 2182.

While some have suggested that the primary significance test is too stringent, particularly for product design cases, we reject any lesser test. The Supreme Court has consistently used "primary significance." As the Court said in a case concerning a product design (the shape of a shredded wheat cereal biscuit), secondary meaning occurs when "the primary significance ... in the minds of the consuming public is not the product but the producer." Kellogg Co. v. National Biscuit Co., 305 U.S. 111, 118, 59 S. Ct. 109, 83 L. Ed. 73 (1938). Where protection of the product design itself is at issue, there is every reason to hold to the usual stringent primary significance test in order to minimize constitutional concerns.

Secondary meaning may be "established in a number of ways," and courts may weigh a number of factors, including the length or exclusivity of use of a mark, "the size or prominence of [plaintiff's] enterprise," and the existence of substantial advertising by plaintiff. President & Trustees of Colby College v. Colby College-New Hampshire, 508 F.2d 804, 807-08 (1st Cir.1975). Other factors include the product's "[e]stablished place in the market" and "[p]roof of intentional copying." 2 McCarthy § 15:30; see also Boston Beer Co. Ltd. Partnership v. Slesar Bros. Brewing Co., 9 F.3d 175, 182 (1st Cir.1993) (listing factors). Secondary meaning may be determined with reference to a particular trade or "branch of the purchasing public." See Colby College, 508 F.2d at 808 (quoting G. & C. Merriam Co. v. Saalfield, 198 F. 369, 373 (6th Cir.1912)) (internal quotation marks omitted).

The district court found that Lund had a likelihood of success of showing that the VOLA had achieved secondary meaning, noting that "there is sufficient evidence in the record ... to conclude that purchasers of high-end bathroom fixtures, including interior designers, now know the VOLA by sight." Lund I, 11 F. Supp. 2d at 121. The district court also found that "[e]nough money and effort has been invested in promoting the VOLA's design that it is recognizable as coming from a unique source, albeit through a variety of distributors," and that "the primary significance of the VOLA design, to the relevant public, is as a high-end product, coming from Lund." Id.

Kohler challenges the district court's determination that the VOLA has most likely acquired secondary meaning. We review the district court's findings of fact for clear error.

The district court's conclusion is doubtful for several reasons. First, as noted, there was no analysis of functionality and there is a relationship between functionality and secondary meaning.

Second, the context suggests otherwise--the district court found, under separate but related doctrines, that the faucets were mostly dissimilar. Cf. 2 McCarthy § 15:38 (stating that copying may give rise to an inference of secondary meaning). The court found that there was no customer confusion. It is in this barren context that the claim is made that something in the design which is not functional serves primarily to signify the source of the faucet and not primarily to signify that it is an aesthetically pleasing faucet.

Third, little is present of the evidence traditionally relied on (but which admittedly is not a sine qua non) for a finding of secondary meaning. The secondary meaning analysis is primarily a subjective one, looking into the minds of potential customers. Customer survey evidence, while not required, is a valuable method of showing secondary meaning. See Boston Athletic Ass'n v. Sullivan, 867 F.2d 22, 32 n. 9 (1st Cir.1989); Colby College, 508 F.2d at 809. As the district court correctly noted, the purported survey evidence was unreliable. The expert testimony, which involved one expert testifying for each side, was contradictory.

Fourth, the VOLA has been sold by a number of different companies in this country and has not been advertised here as coming from Lund. This situation is akin to the use of different marks to identify a single product. "When a product ... design is sold by the authority of plaintiff under several different word marks ..., it is more difficult for plaintiff to prove acquisition of secondary meaning--that is, that the shape or design identifies a single source."8 1 McCarthy § 8:14 (emphasis in original).

Finally, while such testimony is entitled to little weight because the focus is on the understanding of prospective purchasers, the testimony is that Lund did not intend in designing the product that the design signify source.

It is true that the VOLA product has been much advertised or featured, but at times without any source attribution or with attribution to different sources.

Lund relies heavily on the testimony of Corbin, a retailer presented as Lund's expert. Corbin stated that the VOLA was "perceived as a high-end product, a very simple and elegant product; one that is generally known by name as either a Kroin faucet or Jacobsen faucet; ... a design icon...." But to say it is a design success is not to say that the primary significance of the design is to signify its source.

In the end we need not resolve the point. Lund lost on the infringement preliminary injunction and for other reasons, even if Lund had shown secondary meaning, it has not shown itself entitled to relief under the FTDA.

Establishing trademark infringement requires a showing that prospective buyers of the product in question--here, high-end faucets--are likely to be confused as to the product's source. See 15 U.S.C. § 1125(a); Two Pesos, 505 U.S. at 769, 112 S. Ct. 2753; TEC Eng'g Corp., 82 F.3d at 545; Purolator, Inc. v. EFRA Distribs., Inc., 687 F.2d 554, 559 (1st Cir.1982); cf. International Ass'n of Machinists & Aerospace Workers v. Winship Green Nursing Ctr., 103 F.3d 196, 201 (1st Cir.1996) ("[T]he law has long demanded a showing that the allegedly infringing conduct carries with it a likelihood of confounding an appreciable number of reasonably prudent purchasers exercising ordinary care."). Establishing infringement also requires a showing that the plaintiff uses, and thus owns, the mark in question, and that the defendant's mark is similar to or the same as the plaintiff's mark. See DeCosta, 981 F.2d at 605. A plaintiff is not entitled to a preliminary injunction on a trademark infringement claim unless it can persuade the district court that it is likely to be able to demonstrate consumer confusion. See WCVB-TV v. Boston Athletic Ass'n, 926 F.2d 42, 44 (1st Cir.1991).9

This court has identified eight factors to be weighed in determining likelihood of confusion:

(1) the similarity of the marks; (2) the similarity of the goods; (3) the relationship between the parties' channels of trade; (4) the relationship between the parties' advertising; (5) the classes of prospective purchasers; (6) evidence of actual confusion; (7) the defendant's intent in adopting its mark; and (8) the strength of the plaintiff's mark.... No one factor is necessarily determinative, but each must be considered.

Boston Athletic Ass'n, 867 F.2d at 29 (quoting Volkswagenwerk Aktiengesellschaft v. Wheeler, 814 F.2d 812, 817 (1st Cir.1987)) (citations and internal quotation marks omitted). The factors are non-exclusive, however, and are not always apt to the particular facts of a case. See Winship Green, 103 F.3d at 201. In addition, the first factor, similarity, "is determined on the basis of the total effect of the designation, rather than a comparison of individual features." Pignons S.A. de Mecanique de Precision v. Polaroid Corp., 657 F.2d 482, 487 (1st Cir.1981) (internal quotation marks omitted).

The district court applied these eight factors and determined that, taken together, the factors weighed against a finding of a likelihood of consumer confusion. The court found that the strength of the VOLA's mark and the similarity of the VOLA and the Falling Water products were factors that favored the plaintiffs, but that the class of prospective purchasers, the channels of trade, the defendant's intent, and the dissimilarity of the VOLA and Falling Water marks all favored the defendants. The court found that the two remaining factors did not weigh in favor of either party.

Lund argues that the district court misapplied the eight factors listed above, saying the court committed four errors. First, Lund argues that the district court accorded consumer sophistication excessive weight, particularly in reaching its determination that the channels of trade and the classes of prospective consumers made confusion less likely. Second, Lund argues that the court erred in its determination that the dissimilarity of the marks weighed against a finding of consumer confusion. Third, Lund argues that the district court erred when it found that Kohler's intention not to copy the VOLA weighed against a finding of consumer confusion. Fourth, Lund argues that the district court ignored evidence of actual consumer confusion.

Review of factual determinations is for clear error. We have carefully considered each of Lund's arguments, and are unpersuaded that the district court's determination as to the probability of success on each of these points was erroneous. However, we clarify two matters.

First, little weight should be given to the determination that Kohler did not intend to copy the VOLA. In Chrysler, this court commented that "[s]trictly, intent, or lack thereof, does not affect the eyes of the viewer." Chrysler, 118 F.3d at 59 n. 3. We added that "[p]roof of bad intent may, psychologically, hurt as an admission," but "[p]roof of good intent does not change appearance." Id. Similarly, in Star Financial Services we commented that "a finding of good faith is no answer if likelihood of confusion is otherwise established." Star Fin. Servs., 89 F.3d at 11 (quoting Colby College, 508 F.2d at 811-12) (internal quotation marks omitted). Thus a finding of no intent to copy may not outweigh other factors that suggest a likelihood of confusion.

Second, the district court's discussion of post-sale confusion was based on an erroneous premise. In finding that post-sale confusion was unlikely, the court commented that "there is little or no chance that [the faucets] will be resold to unwary consumers." Lund I, 11 F. Supp. 2d at 123. Post-sale confusion refers not to the resale of the original product, however, but to the risk that non-purchasers, who themselves may be future consumers, will be deceived. See 3 McCarthy § 23:7 (noting that "[t]he damage to the senior user ... is that consumers could acquire the prestige value of the senior user's product by buying the copier's cheap imitation," and that in such a case, "[e]ven though the knowledgeable buyer knew that it was getting an imitation, viewers would be confused"). For example, in a case involving counterfeit Rolex watches, the court commented that

[i]ndividuals examining the counterfeits, believing them to be genuine Rolex watches, might find themselves unimpressed with the quality of the item and consequently be inhibited from purchasing the real time piece. Others who see the watches bearing the Rolex trademarks on so many wrists might find themselves discouraged from acquiring a genuine because the items have become too common place and no longer possess the prestige once associated with them.

Rolex Watch U.S.A., Inc. v. Canner, 645 F. Supp. 484, 495 (S.D.Fla.1986). In other words, even if the purchaser of a Falling Water faucet knew that she was buying the Kohler faucet and not the VOLA, the district court could still find a likelihood of confusion if subsequent viewers of the faucet would believe it to be a VOLA. Cf. Keds Corp. v. Renee Int'l Trading Corp., 888 F.2d 215, 222 (1st Cir.1989) (noting that "point of sale confusion was not the only issue" because "prospective consumers, viewing the clothes on other people, would be confused as to the origin of the goods"). However, because the district court's determination of no likelihood of post-sale confusion does not appear to have been a key factor in its decision, we do not believe it undercuts the court's conclusion regarding probability of success on the merits.

The district court correctly found that Lund was unlikely to prevail on the merits of its infringement claim, and so correctly denied the preliminary injunction on those grounds.

Lund obtained the preliminary injunction against Kohler's distribution and promotion of the Falling Water faucet based on the district court's finding of likelihood of success under the new federal anti-dilution statute, the FTDA, which became effective in 1996. The injunction rested on the conclusion that Lund had established a likelihood of success of showing two essential elements. The first is that the "mark," that is, the VOLA product design as an identifying mark, was "famous." The second is that Kohler's Falling Water faucet "diluted" Lund's mark. Both the terms "famous" and "dilution" are terms of art given specific rigorous meanings by the FTDA.

We start with the language of the Act. The FTDA provides:

The owner of a famous mark shall be entitled, subject to the principles of equity and upon such terms as the court deems reasonable, to an injunction against another person's commercial use in commerce of a mark or trade name, if such use begins after the mark has become famous and causes dilution of the distinctive quality of the mark, and to obtain such other relief as is provided in this subsection.

15 U.S.C. § 1125(c)(1).

The FTDA created a new "federal cause of action to protect famous marks from unauthorized users that attempt to trade upon the goodwill and established renown of such marks and, thereby, dilute their distinctive quality." H.R.Rep. No. 104-374, at 3, reprinted in 1995 U.S.C.C.A.N at 1030. Congress acted against a "patch-quilt system" of state law protection in which approximately twenty-five states had laws prohibiting trademark dilution. Id. While creating a federal cause of action, Congress expressly did not preempt state law, but attempted to create uniformity through the availability of a federal cause of action. A second consideration--protection of famous marks of U.S. companies abroad--also motivated Congress. See id. at 4, reprinted in 1995 U.S.C.C.A.N at 1031. Enactment of the FTDA was consistent with agreements which were part of the Uruguay Round of the General Agreement on Tariffs and Trade (specifically, the Agreement on Trade-Related Aspects of Intellectual Property Rights, including Trade in Counterfeit Goods ("TRIPS")) and the Paris Convention. Enactment of the law was also thought to be of value to the U.S. in bilateral and multi-lateral trade negotiations. See id.

Sponsors of the bill articulated the type of problem the Act was meant to solve:

[T]his bill is designed to protect famous trademarks from subsequent uses that blur the distinctiveness of the mark or tarnish or disparage it, even in the absence of a likelihood of confusion. Thus, for example, the use of DuPont shoes, Buick aspirin, and Kodak pianos would be actionable under this bill.

141 Cong. Rec. S19306, S19310 (daily ed. Dec. 29, 1995) (statement of Sen. Hatch). Thus the archetypal problems involved non-competing products as to which there could, by definition, be no confusion and a world-famous brand name which was either tarnished or blurred by its application to a different product which was obviously trading on the good will of that name.

The language of the FTDA itself is, however, not limited to addressing these archetypal problems. A few observations are in order. First, the Act applies to products which are competitors (as is true of the faucets here) as well as to products which are totally dissimilar and are not competitors. Second, the Act applies to a famous "mark" and does not restrict the definition of that term to names or traditional marks. In the absence of such a restriction, the Act applies to all types of marks recognized by the Lanham Act, including marks derived from product designs. Kohler's argument that the Act cannot, as a matter of statutory interpretation, be applied to product design is rejected. Third, the Act applies even where there is no customer confusion, a point with consequences discussed later. Fourth, the additional protection afforded by the Act requires that a mark go beyond what is required for ordinary Lanham Act protection. Only those marks which are "distinctive and famous" are protected. 15 U.S.C. § 1125(c)(1). As set forth in the 1987 Trademark Review Commission Report, the precursor of the 1996 Act, this language reflected "the policy goal that to be protected, a mark had to be truly prominent and renowned." 3 McCarthy § 24.91 (citing The United States Trademark Association Trademark Review Commission Report and Recommendations to USTA President and Board of Directors, 77 Trademark Rep. 375, 459-60 (1987)). With this background, we turn to the district court's two essential conclusions.

The requirements of commercial use, non-functionality, and distinctiveness are common to both the infringement claim and the FTDA dilution claim. In order to fall within the sphere of protection against dilution the FTDA adds an additional requirement: the FTDA grants protection only to famous marks. See 15 U.S.C. § 1125(c); cf. Gilson § 5.12[a] (calling the FTDA "a major breakthrough for an elite category of trademark owners"). The FTDA provides a non-exclusive list of eight factors that courts should consider in determining whether a mark is "distinctive and famous." 15 U.S.C. § 1125(c)(1). The eight statutory factors are:

(A) the degree of inherent or acquired distinctiveness of the mark;

(B) the duration and extent of use of the mark in connection with the goods or services with which the mark is used;

(C) the duration and extent of advertising and publicity of the mark;

(D) the geographical extent of the trading area in which the mark is used;

(E) the channels of trade for the goods or services with which the mark is used;

(F) the degree of recognition of the mark in the trading areas and channels of trade used by the marks' owner and the person against whom the injunction is sought;

(G) the nature and extent of use of the same or similar marks by third parties; and

(H) whether the mark was registered under the Act of March 3, 1881, or the Act of February 20, 1905, or on the principal register.

Id. The district court found likelihood of success on the claim that the VOLA's design was a famous mark. Kohler challenges this finding.

Both the text and legislative history of the original bill in 1988 and the FTDA itself indicate a congressional intent that courts should be discriminating and selective in categorizing a mark as famous. For example, the Senate Judiciary Committee Report on the 1988 precursor bill10 said the Committee wished "to underscore its determination that the new dilution provisions should apply only to ... very unique marks." 3 McCarthy § 24:92 n. 6 (quoting S.Rep. No. 100-515, at 41-42) (internal quotation marks omitted). The Trademark Review Commission noted that the showing of fame required employment of a "higher standard" than fame among an "appreciable number of persons" in order to be "eligible for this extraordinary remedy." Id. n. 8 (quoting The United States Trademark Association Trademark Review Commission Report, 77 Trademark Rep. at 461) (internal quotation marks omitted). One commentator has referred to this category of famous marks as "Supermark[s]." See Gilson § 5.12[a]. As the Restatement (Third) of Unfair Competition notes:

A mark that evokes an association with a specific source only when used in connection with the particular goods or services that it identifies is ordinarily not sufficiently distinctive to be protected against dilution.

Restatement (Third) of Unfair Competition § 25 cmt. e (1995).

The record here reflects that there was not sufficient attention paid to the heightened fame standard that the FTDA establishes. The district court found that "in its market, the VOLA's design is famous and distinctive" based on its prior analysis under the infringement claim that "the VOLA mark is strong, and has acquired secondary meaning in its market; and it has been used for some twenty years in this country." Lund I, 11 F. Supp. 2d at 125. Both the Restatement (Third) of Unfair Competition and the state anti-dilution statutes against the background of which Congress enacted the FTDA make clear that the standard for fame and distinctiveness required to obtain anti-dilution protection is more rigorous than that required to seek infringement protection. See Restatement (Third) of Unfair Competition § 25 cmt. e. As one commentator has stated:

Certainly, the mere acquisition of secondary meaning to achieve trademark status in a non-inherently distinctive designation is nowhere near sufficient to achieve the status of "famous mark" under the anti-dilution statute. The acquisition of secondary meaning merely establishes the minimum threshold necessary for trademark status: section 43(c) requires a great deal more.

3 McCarthy § 24:91; see also Gilson § 5.12[c][ii] ("A trademark can certainly be distinctive without being famous, but it cannot be famous without being distinctive.").

We do not understand the district court's conclusion about fame, despite some ambiguity, to have rested solely on its conclusion that the VOLA faucet was distinctive (because it had acquired secondary meaning); any such per se analysis would be erroneous. But it appears that the district court did not apply the more rigorous definition of fame under the FTDA. While one can posit a case having very strong facts which support a distinctiveness finding based on secondary meaning and then using those facts to support the separate, more rigorous analysis of fame, the facts in this case do not have such strength. Lund has a difficult case to establish fame through the product design of the VOLA faucet, and this record is not strong. There has been no use by Kohler of the VOLA name, or any other name by which the faucet is known. Kohler's use, if any, is of a like product design. There is little to suggest that this product design, itself unregistered and not inherently distinctive, is so strong a mark and so well publicized and known that it has achieved the level of fame Congress intended under the Act. Consumer surveys could be used as evidence of such fame, but consumer surveys are absent from this case. Additionally, although some marks, such as COCA-COLA, may be so famous as to be judicially noticed, see Gilson § 5.12[c][iii], the VOLA faucet is far from being a candidate for such judicial notice.

Further, national renown11 is an important factor in determining whether a mark qualifies as famous under the FTDA. Although the district court found that "in the world of interior design and high-end bathroom fixtures, the VOLA is renowned," Lund I, 11 F. Supp. 2d at 126, and that the faucet has been featured and advertised in national magazines and displayed in museums, whether the VOLA's identifying design is sufficiently famous to qualify for the FTDA's protection is far from clear. In light of the rigorous standard for fame, we find that Lund has not met its burden of showing likelihood of success.

Under the FTDA, even if a mark is famous there is no relief unless that mark has been diluted. As the district court noted, there are two types of dilution recognized: blurring and tarnishing. This case involves no claim of tarnishing, an area in which Congress expressed a strong interest. Further, in light of the finding of no customer confusion, only a particular type of blurring may be involved.

The intellectual origins of the dilution doctrine are traced to a 1927 Harvard Law Review article, which urged protection against "the gradual whittling away or dispersion of the identity and hold upon the public mind of the mark or name by its use upon non-competing goods." Schecter, The Rational Basis of Trademark Protection, 40 Harv. L.Rev. 813, 825 (1927). Although the origins of the doctrine are concerned with non-competing goods, Congress used language in the FTDA which extends dilution protection even to competing goods. "Dilution" is defined as "the lessening of the capacity of a famous mark to identify and distinguish good or services." 15 U.S.C. § 1127; see also H.R.Rep. No. 104-374, at 3, reprinted in 1995 U.S.C.C.A.N at 1030 (stating that dilution "applies when the unauthorized use of a famous mark reduces the public's perception that the mark signifies something unique, singular, or particular"); 3 McCarthy § 24:93 ("The crux is whether this particular challenged use lessens the capacity of the famous mark to carry out its role as a trademark--namely, to identify and distinguish.").

This case differs in several respects from usual dilution cases. First, unlike most claims of dilution by tarnishment or blurring, the aspect of its product that Lund seeks to protect--the design of the VOLA faucet--likely could have been protected by a design patent.12 The possibility of obtaining a design patent is not dispositive of the availability of trade dress protection: more than one form of intellectual property protection may simultaneously protect particular product features. Moreover, design patent protection would not have provided protection identical to that sought here. Nevertheless, the availability of design patent protection does suggest that the claim in this case differs fundamentally from the claims the drafters of the FTDA had in mind--cases where dilution protection is the only form of protection available for a famous mark threatened by unauthorized use of the mark that lessens the mark's capacity to identify its source. Congress's intent to provide protection where none previously existed is evidenced by the House Report's statement that "[a] federal dilution statute is necessary because famous marks ordinarily are used on a nationwide basis and dilution protection is currently only available on a patch-quilt system of protection, in that only approximately 25 states have laws that prohibit trademark dilution." H.R.Rep. No. 104-374, at 3, reprinted in 1995 U.S.C.C.A.N. at 1030.

Second, Lund is seeking protection against a direct competitor. Although the FTDA states that dilution protection is available against unauthorized use of a famous mark that lessens the mark's ability "to identify and distinguish goods or services, regardless of the presence or absence of ... competition between the owner of the famous mark and other parties," 15 U.S.C. § 1127 (emphasis added), dilution protection has most often been extended to non-competing uses of a mark, see, e.g., H.R.Rep. No. 104-374, at 3, reprinted in 1995 U.S.C.C.A.N. at 1030 (giving examples of "DUPONT shoes, BUICK aspirin, and KODAK pianos"); 3 McCarthy § 24.72 (noting "split of authority under ... state statutes as to whether the anti-dilution rule is applicable where the parties are in competition"). The FTDA recognizes the possibility that dilution may occur in some circumstances where, although some consumers are not confused as to the products' sources, a competitor's use of a mark tarnishes or blurs a senior mark. Nevertheless, such cases are likely to be exceptions to the more common cases of dilution by non-competing marks. Dilution laws are intended to address specific harms; they are not intended to serve as mere fallback protection for trademark owners unable to prove trademark infringement.

While there may be a tendency to think of dilution in terms of confusion, Congress made it clear that dilution can occur even in the absence of confusion. It is simple to see why that should be so when non-competing goods are at issue. No one would confuse Kodak pianos with Kodak film, but the use of the name on the piano could dilute its effectiveness as a mark for the film. But Congress did not say that there can be dilution without confusion only among non-competing goods. As the district court aptly noted, the analysis becomes complicated when the concept of blurring is applied to competing similar products.

We deal first with the approach taken by the district court, an approach which had support in precedent. The district court articulated the standard for determining blurring as follows: "Lund must demonstrate that 'the use of a junior mark has caused a lessening of demand for the product or services bearing the famous mark.' " Lund I, 11 F. Supp. 2d at 126 (quoting Ringling Bros.-Barnum & Bailey Combined Shows, Inc. v. Utah Div. of Travel Dev., 955 F. Supp. 605, 616 (E.D.Va.1997)). This, we think, is not the correct standard. As Kohler observes, demand for one product is almost always lessened whenever a competing product achieves a measurable degree of success. Further, blurring has to do with the identification of a product and that is not the same thing as a lessening of demand.

In addressing the dilution claim, the district court used the "Sweet factors," named after the six factors set forth in Judge Sweet's concurrence in Mead Data Central, Inc. v. Toyota Motor Sales, U.S.A., Inc., 875 F.2d 1026, 1035 (2d Cir.1989) (Sweet, J., concurring), which involved a claim brought under a New York dilution statute. Some district courts have used these factors to examine whether dilution exists under the FTDA. See, e.g., Ringling Bros.-Barnum & Bailey Combined Shows, Inc. v. B.E. Windows Corp., 937 F. Supp. 204, 211-14 (S.D.N.Y.1996); WAWA Dairy Farms v. Haaf, 40 U.S.P.Q.2d (BNA) 1629, 1632-33 (E.D.Pa. Aug.7, 1996), aff'd, 116 F.3d 471 (3d Cir.1997); Clinique Laboratories, Inc. v. Dep Corp., 945 F. Supp. 547, 562 (S.D.N.Y.1996).

Kohler argues that Judge Sweet's six-factor test is inappropriate in determining whether dilution has occurred for purposes of the FTDA. We agree. The Sweet factors have been criticized by both courts and commentators for introducing factors that "are the offspring of classical likelihood of confusion analysis and are not particularly relevant or helpful in resolving the issues of dilution by blurring." 3 McCarthy § 24:94.1. The six Sweet factors are: "(1) similarity of the marks (2) similarity of the products covered by the marks (3) sophistication of consumers (4) predatory intent (5) renown of the senior mark [and] (6) renown of the junior mark." Mead Data, 875 F.2d at 1035 (Sweet, J., concurring). McCarthy urges that only the first and fifth of Judge Sweet's factors--the similarity of the marks and the renown of the senior mark--are relevant to determining whether dilution has occurred. See 3 McCarthy § 24:94.1; see also Hershey Foods Corp. v. Mars, Inc., 998 F. Supp. 500, 520 (M.D.Pa.1998) (stating that "whether the products are similar or not adds nothing to the analysis" because "dilution can apply to competitors"); Klieger, Trademark Dilution: The Whittling Away of the Rational Basis for Trademark Protection, 58 U. Pitt. L.Rev. 789, 826-27 (1997) (noting that "few of these factors bear any relation to whether a particular junior use will debilitate the selling power of a mark" and that "[s]o long as a mark qualifies for dilution protection and the senior and junior uses of the mark are not so unrelated as to foreclose the possibility of a mental connection, blurring ... is a foregone conclusion"). These criticisms are well taken.

The district court's finding of likelihood of dilution by blurring depended on its use of inappropriate Sweet factors. As McCarthy points out, use of factors such as predatory intent, similarity of products, sophistication of customers, and renown of the junior mark work directly contrary to the intent of a law whose primary purpose was to apply in cases of widely differing goods, i.e. Kodak pianos and Kodak film. See 3 McCarthy § 24:94.1.

There are difficulties with the Sweet factors even when used with competing goods. Blurring occurs in the minds of potential customers. Predatory intent tells little about how customers in fact perceive products. That customers are sophisticated may well mean less likelihood of blurring. That customers knowingly choose to pay less to get a similar product, and trade lower price against having a product of greater fame, does not, contrary to Lund's argument, establish blurring. Indeed, the district court's findings, in the infringement context, of dissimilarity and sophistication of the customers tend to cut against any finding of blurring. "The familiar test of similarity used in the traditional likelihood of confusion test cannot be the guide [for dilution analysis], for likelihood of confusion is not the test of dilution." Id. § 24:90.1. Instead, the inquiry is into whether target customers are likely to view the products "as essentially the same." Id.

There is a more fundamental problem here in attempting to apply the dilution analysis to the design itself of the competing product involved. We doubt that Congress intended the reach of the dilution concept under the FTDA to extend this far and our doubts are heightened by the presence of constitutional constraints. Where words are the marks at issue it is easy to understand that there can be blurring and tarnishment when there is a completely different product to which the words are applied. The congressional history described earlier gives such examples. What is much more difficult is to see how dilution is to be shown where some of a design is partially replicated and the result is largely dissimilar and does not create consumer confusion. If that is so, as is true here, then it is difficult to see that there has been dilution of the source signaling function of the design (even assuming that such a function has been established through secondary meaning).

Instead, it appears that an entirely different issue is at stake--not interference with the source signaling function but rather protection from an appropriation of or free riding on the investment Lund has made in its design. That investment is usually given protection by patents, which have a limited duration. See W.T. Rogers Co., 778 F.2d at 348. But again, that free riding or appropriation appears to be of the beauty of the object and not of the source, and may in fact be good for consumers. As the district court observed, these sophisticated buyers know a different source is involved and are, accordingly, most likely purchasing because of the aesthetics. And even if there is some appropriation or free riding, the extent of it here is not clear. Certainly it is not plausible to think that Congress intended to protect aesthetic characteristics by simply assuming harm or damages based on the fact that the plaintiff will sell less if the defendant sells more. What is clear is that the interests here are not the interests at the core of what Congress intended to protect in the FTDA.

It is possible that Congress did not really envision protection for product design from dilution by a competing product under the FTDA, but the language it used does not permit us to exclude such protection categorically and rare cases can be imagined. But a broad reading of dilution would bring us close to the constitutional edge, and we decline to attribute such brinksmanship to Congress, and so insist on rigorous review.

Under the interpretation of the fame and dilution requirements for the FTDA set forth today, the requirements for granting the preliminary injunction have not been met.

Kohler argues that the FTDA may never constitutionally be applied to enjoin a competitor in a product design trade dress case. Kohler's constitutional challenge involves two steps. First, Kohler argues that applying the FTDA to product designs grants patent-like protections for an unlimited period of time. Second, Kohler argues that Congress's Commerce Clause power--the basis of Congress's regulation of trademarks and trade dress--cannot be used to trump the Patent Clause. Lund responds that federal anti-dilution legislation is fully consistent with Congress's Commerce Clause power, and that patent and trademark laws protect different interests and serve different goals.

The district court correctly noted that Kohler faced a "very high preliminary injunction standard on their [constitutional] claim," and that the statute was "presumptively constitutional." Lund II, 11 F. Supp. 2d at 134. The court concluded that defendant Kohler was unlikely to succeed with its argument that the FTDA is unconstitutional in all product design contexts.

Kohler's constitutional attack on application of the FTDA here is mooted by our resolution of the injunction issue. To the extent Kohler is mounting something akin to a facial attack, we think it better not to address constitutional issues in the abstract. The resolution of any conflict between the Patent Clause and the FTDA is better handled on specific facts which present the issues with clarity, and not on the basis of theoretical impacts.

Both the parties and the district court labored through this case without the benefit of binding precedent on a number of key and difficult issues, particularly the interpretation of the FTDA. The denial of the preliminary injunction on the infringement claim is affirmed. The grant of the injunction on the FTDA claim is vacated inasmuch as there were no findings on functionality and standards for determining both fame and dilution under the FTDA were used which are different from those announced today and the evidence does not show probability of success under those standards. The case is remanded for further proceedings not inconsistent with this opinion. No costs are awarded.

BOUDIN, Circuit Judge, concurring.

Ordinarily the creator of something new--a useful device, a pharmaceutical drug, an ornament, a painting--owns any such object that he or she makes but can prevent its replication by others only pursuant to the patent and copyright laws. A central limitation on patent and copyright protection, stemming from the Constitution itself, is that it is limited in time. See U.S. Const. art. I, § 8, cl. 8; 35 U.S.C. §§ 101, 154 (20-year utility patent); 35 U.S.C. §§ 171, 173 (14-year design patent); 17 U.S.C. §§ 302-304 (various limited-time copyrights). After the period of protection expires, copying by others is allowed. This case presents, in addition to other problems, the threat that the permanent injunction sought may prevent the Lund faucet from being duplicated forever.

In substance, Lund created a faucet configuration called VOLA that involves a downward curving water pipe protruding from a wall (rather than the sink itself) and a similarly protruding control rod to regulate both water flow and temperature (instead of the usual pair of spigots). The concept of a wall-based mechanism, and the size and shape and relative proportions of the pipe and control rod, are pleasing to customers. The resulting VOLA faucet configuration is successful and recognized in the trade.

Lund has now obtained a preliminary injunction against Kohler forbidding Kohler from selling Kohler's own faucet system that embodies the same concept and a similar but not identically shaped pipe and control rod. What Lund regards as the scope of its "monopoly" is not entirely clear: all we know is that this Kohler faucet system has been enjoined. What will be assumed, or else the legal problems would be even greater than they are, is that the concept of a water pipe and control rod protruding from a wall is not sought to be prohibited but only one whose shape and size are similar to VOLA.1 Even so, a serious problem exists because the assumed protection (if the preliminary injunction matures into a permanent one) could be perpetual.

Lund says that this perpetual protection of its design is permitted because its source is not a design patent2--Lund has not sought one--but rather the trademark laws. Trademarks are ordinarily conceived as names (e.g., Kodak) or marks (the AT & T striped circle logo), but trademarks also include so-called "trade dress," see 15 U.S.C. § 1127; and this in turn includes the ornamental design of an article where the design performs the traditional "source signaling" function of trademarks. Usually, this is achieved by the "wrapping" (e.g., the classic Coca Cola bottle) or some ornament affixed to the product (the star symbol on the Mercedes hood); but the design of the product itself can sometimes qualify.3

A permanent restriction on copying the design of a product might still seem offensive to patent and copyright policy, but at least this protection can be explained where the replication would otherwise confuse buyers as to source, traditionally the concern of trademark protection. The difficulty in this case arises because Congress has removed the requirement of confusion for protection of "famous" trademarks and said that, in the alternative, they will be protected against "dilution" even where no confusion exists as to source. 15 U.S.C. § 1125(c). Congress was worried that even without customer confusion as to source a famous trade name or mark, like "Kodak," could be blurred or degraded by being attached by others to their own products (e.g., sports clothing, dog food). See 141 Cong. Rec. S19306, S19310 (daily ed. Dec. 29, 1995) (statement of Sen. Hatch).

By contrast to a trademark consisting of a name or insignia, a product design will not often qualify as a trademark inviting protection from dilution. A trade name or device usually identifies source, but the design of the product itself in most cases makes the product more efficient, attractive, or both. If more efficient, trademark protection is foreclosed altogether under the "functionality" doctrine, see Qualitex, 514 U.S. at 164, 115 S. Ct. 1300; Two Pesos, 505 U.S. at 775, 112 S. Ct. 2753; and if attractiveness is the primary function, the product design does not normally qualify as a candidate to achieve secondary meaning and thus trademark status. See Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844, 851 n. 11, 102 S. Ct. 2182, 72 L. Ed. 2d 606 (1982).

But conceivably in some rare case a product design might have source signaling as its main function (perhaps the old 1950's Cadillac tail fin might be an instance) and be copied by a competitor (say, for Volkswagen Beetles) without any risk of customer confusion. At that point, if the new trademark statute is taken literally, Cadillac might then claim a permanent monopoly on its tail fin design, subject only to a showing that Volkswagen's use threatened to "dilute" the strength of that tail fin as a Cadillac trademark. See 15 U.S.C. § 1125(c). One may wonder whether Congress ever contemplated such a result, but the statute's language arguably permits it or at least does not explicitly foreclose the possibility.

If so, much turns on how dilution is defined and proved. The federal statute is recent, but state laws have protected against dilution for some time, and under those statutes there is some tendency to assume that dilution is threatened wherever the famous trademark is copied.4 After all, the concern is with the long-term "whittling away" of the strength of the mark by other unauthorized uses, see L.L. Bean, Inc. v. Drake Publishers, Inc., 811 F.2d 26, 30 (1st Cir.1987), and, absent confusion, there may be no immediate damage from any single use. Where only a word or symbol is forever preempted, this protective approach toward the trademark may make sense, and in all events does not pose much risk to the policies of the patent and copyright clauses.

The situation is quite different where design of an article is given permanent protection without any threat of confusion. If buyers prefer articles of that design, then the public pays a price whenever the design cannot eventually be used by other manufacturers. In the case of patents and copyrights, the foreclosure of competition is deemed a price worth paying for a limited time--in order to induce such creations--but only for a limited time. Is this policy of time-limited protection, constitutional at its core, overcome wherever dilution is threatened--even though no confusion can be proved and any impairment of the trademark may be only potential and directed primarily to protecting the seller rather than the public?

One might point to the possible diversion of sales from Lund to Kohler as practical harm in this very case. Yet such diversion does not prove either confusion or dilution: a copied product can easily take sales from the original manufacturer if the copied product is better made or sold at a lower price or the design is slightly better--outcomes generally beneficial to the public. Giving weight to diversion of sales, in the absence of confusion of customers, simply underscores that the protection is generally similar to that afforded for patented articles--save that the protection here may be unlimited in time. Giving patent-like protection a new name does not avoid constitutional limitations. Cf. Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141, 146, 109 S. Ct. 971, 103 L. Ed. 2d 118 (1989).

The traditional interest in trademark protection is stretched very thin in dilution cases where confusion is absent and a professed aim is to protect the maker's investment in the trademark. See H.R.Rep. No. 104-374, at 3 (1995), reprinted in 1995 U.S.C.C.A.N. at 1030. And the threat to the public interest, ordinarily countered by the time limit on patent protection, is acute where a permanent protection is offered not to a word or symbol but to the design of an article of manufacture. In such an instance, it is a difficult constitutional question whether protecting the investment can outweigh the public interest in replication, a question best deferred unless and until all other preconditions for protection are resolved in favor and decision on this last issue is absolutely necessary. See El Dia, Inc. v. Hernandez Colon, 963 F.2d 488, 494 (1st Cir.1992).