This is a trade dress infringement action brought under section 43(a) of the Lanham Act, 15 U.S.C.A. Sec. 1125(a) (West Supp.1994). Plaintiff Duraco Products, Inc. ("Duraco") appeals from an order of the district court denying its motion for a preliminary injunction against defendants Joy Plastic Enterprises, Ltd. ("Joy"), d/b/a Backyard Products, and Travis Products, Inc. ("Travis"). Duraco, a manufacturer of plastic planters for use in gardens, claims that Joy has infringed the trade dress of Duraco's most popular product by marketing a planter with a similar shape and texture, and that Travis is liable for manufacturing the molds for Joy's planter. Because Duraco's claim is predicated upon infringement of the trade dress of the product itself, the appeal requires us to confront a difficult area of trade dress law--that dealing with the circumstances under which product configurations, in contrast to product packaging, can, in Lanham Act parlance, constitute inherently distinctive trade dress thus serving as a designator of origin that will protect the plaintiff's product design features against copying.

We conclude that traditional trade dress doctrine does not "fit" a product configuration case because unlike product packaging, a product configuration differs fundamentally from a product's trademark, insofar as it is not a symbol according to which one can relate the signifier (the trademark, or perhaps the packaging) to the signified (the product). In other words, the very basis for the trademark taxonomy--the descriptive relationship between the mark and the product, along with the degree to which the mark describes the product--is unsuited for application to the product itself.

However, we also think that there is a proper set of circumstances for treating a product configuration as inherently distinctive. These circumstances are characterized by a high probability that a product configuration serves a virtually exclusively identifying function for consumers--where the concerns over "theft" of an identifying feature or combination or arrangement of features and the cost to an enterprise of gaining and proving secondary meaning outweigh concerns over inhibiting competition, and where consumers are especially likely to perceive a connection between the product's configuration and its source. We conclude that, to be inherently distinctive, a product feature or a combination or arrangement of features, i.e., a product configuration, for which Lanham Act protection is sought must 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 district court applied a different standard, and in the ordinary course we might remand for reconsideration under the proper test. However, our examination of the record persuades us that, under the standard we adopt, no factfinder could reasonably conclude that Duraco has demonstrated a likelihood of success on the merits by meeting the threshold distinctiveness requirement of section 43(a) either through a showing of inherent distinctiveness or, failing that, by establishing secondary meaning. We will therefore affirm the district court's order. However, the district court will have to conduct a final hearing at which it will apply the newly announced standard. In view of our disposition, we need not reach the other grounds that the district court gave for its denial of a preliminary injunction, i.e., functionality of the trade dress and failure to show a likelihood of confusion.

Duraco's most profitable products, its "Grecian Classics" plastic planters, account for one tenth of its nearly $35 million in annual sales. These planters, shaped like a Grecian urn, are made in two sizes, with diameters of twelve and eighteen inches and heights of ten and fifteen inches, respectively. Their plastic construction makes them inexpensive and durable. But the key to their considerable success, according to Duraco, is that a careful combination of ornamental features creates in them the illusion of marble, cement, or stone construction.

The stimulus for the Duraco urn's design was a suggestion by Robert Armstrong, a Senior Buyer at K-Mart, Duraco's largest retailer customer. Armstrong had run across urn-shaped planters similar in appearance to the eventual design of the Grecian Classics at a trade show in the Federal Republic of Germany sometime in 1984. Realizing that like products were not then available in the United States, Armstrong met with Duraco officials to describe his fortuitous discovery and to encourage Duraco to manufacture such an item. Duraco, in turn, set about to satisfy Armstrong's interest. It surveyed Grecian urns at statuary stores and explored its own archives. As it happened, in the late 1970s Duraco had tried to market the "Cotswold Planter," an English-made Grecian plastic planter. Poor sales, perhaps attributable to its relatively high retail price tag--$14.99 compared to under $5.00 for the Grecian Classics--caused Duraco to drop the product two years later. The Cotswold may have survived in United States commerce for some time thereafter, but was no longer available at the time of the Armstrong-Duraco conference.

The Grecian Classics are much like, but not clones of, the Cotswold. Both planters have an hourglass-like design and fluting, though the Cotswold has a higher base (hence a higher center of gravity), softer lines, and a less realistic texture. Despite the differences, Armstrong would have been pleased with a replica of the Cotswold urn.

Once Duraco showed Armstrong its prototype, K-Mart committed itself to purchase 100,000 Grecian Classics planters from Duraco. With no competing plastic urn in the United States sporting a similar, detailed design, sales of the eighteen-inch Grecian Classics planter reached 460,000 the first year, making it Duraco's leading seller. This overwhelming success persuaded Duraco to manufacture a twelve-inch cousin; sales continued to soar.

Today, Duraco markets the bulk of its planter wares directly to large retailers, primarily large discount department stores; some distributors are directly supplied. Advertising is typically done cooperatively with the retailers: the parties share costs, and Duraco allows the retailer to use two percent of sales receipts for advertising, redeemable in cash or credit. The media generally relied upon for consumer advertising are Sunday newspaper fliers, magazines, circulars, and newspaper ads. In those types of advertising, the planters are generally depicted in juxtaposition with other outdoor garden products, such as plant food and watering cans. Duraco also solicits retailers with brochures and trade advertisements.

In its marketing endeavors, Duraco encourages, but does not contractually require, its retailer customers to place either the Duraco tradename or its registered trademark "Garden Scene" logo in their advertisements. Although a retailer's failure to display Duraco's logo results, at worst, in an admonition, half of the planter advertisements contain the logo. Nevertheless, brand-name awareness in the outdoor garden products area is slight--less than 0.5 percent of consumers in Duraco's survey recognized the Duraco name. This is probably so because planters are inexpensive and are typically bought as an impulse item.

A number of enterprises compete with Duraco in the plastic planter market, amongst them the defendant Joy. Joy, also known by its "Backyard Products" logo, calls its planters "Ultimate Urns." Defendant Travis manufactures the molds for the plastic planters; Joy markets them. The interest of Joy's President Thomas Gay in producing a planter configured like a classic Grecian urn had been piqued by a color brochure depicting the Cotswold urn. In 1987 and 1988, he approached buyers at K-Mart and another retailing chain, both of which already stocked Duraco's Grecian Classics, to elicit information regarding the potential for a competitor to Duraco in the plastic urn market. Having received favorable feedback, Gay undertook the development of the Backyard Products urn.

During the development phase Gay had access to various urns in the market, including the Grecian Classics. His considered observation and analysis of all those urns' features convinced him to design one with a deeper bowl--to hold more soil and water for enhanced root development--and with a lower center of gravity--because some other urns, such as the Duraco urns, were too top-heavy and subject to tipping over. After completing his design, Gay engaged defendant Travis to create the plastic injection mold for the Ultimate Urns. Joy's Ultimate Urns are strikingly similar in appearance to Duraco's Grecian Classics. Both planters have a similar two-part construction consisting of a top "bowl" section that connects at a joint with the "base" or "pedestal" of the planter. Like Duraco, Joy offers a twelve-inch, reduced-scale version of its eighteen-inch model.

According to Richard Husby, Duraco's Vice President of Marketing and Sales, the Ultimate Urns cannot truly compare to the Grecian Classics by virtue of Joy's failure to match Duraco's punctilious quality control. Joy's planters, of comparatively poor quality according to Husby, can be found in the market with drooping flashing, poor color, and sharp edges and ridges. The inferior coloring is said to stem from Joy's reliance on a less expensive coloring process, and Husby testified that Joy's inferior molding technique causes blotching on the urns' bottoms. The uneven edges of Joy's planters are caused by Joy's cutting away, rather than sanding off, the flashing.1

Other differences between the Grecian Classics and Ultimate Urns, unrelated to quality, can be noted, though only in a sharply focused side-by-side comparison. The Duraco planter has higher fluting (used on the bowl and pedestal of the planters to create the illusion of an actual marble or alabaster planter) and a wider "landing" on the lip between each flute; the Backyard Products urn has a broader base, a lower center of gravity (causing it to withstand tipping better), and a twenty percent greater fill capacity. The side of the Backyard Products urn's bowl is straight-edged, whereas the Grecian Classics' curves in an arc. Visual coincidence in detail can be noted in the "egg and dart" patterns, used primarily for decoration and subsidiarily to strengthen the rim, on the lips of the planters' bowls.

Duraco inadvertently became aware of Joy's competition when Duraco's President and Chief Executive Officer John Licht stumbled across a Backyard Products planter while shopping in a New Jersey retail store. He saw a defective urn--one side of it was flattened, apparently because it had been removed from the mold and laid on its side in a shipping carton while still hot--and purchased it, thinking it to be a defective Duraco urn. He turned the urn over to Duraco's vice president, demanding to know how a defective urn could have slipped through Duraco's quality control safety net into the marketplace. When the vice president returned with news that the urn was Joy's, not Duraco's, Licht phoned Duraco's attorney.

Duraco claims that its "trade dress" differs significantly from the designs of all other competitors save Joy. Duraco does not claim that any of the classical elements in its urn's design is protectable; rather, Duraco defines the uniqueness of its trade dress according to the total composition of all the elements of its urns' configuration, including the rim, the finish, the joining of the top and bottom halves, and the color results. Duraco not only adduced evidence that other competitors had come up with patently different designs for plastic planters, but also presented five designs by its expert designer that, according to the expert, could compete effectively in the market for "contemporary interpretations of neoclassic planters using modifications of classical motifs." Mem. op. at 9. His five designs assertedly achieved the desired motif through variations on sundry classical elements, including the fluting, rims, radii, pedestals, and egg-and-dart design on the bowl's rim. Thus, in Duraco's submission, Joy could have competed effectively without infringing Duraco's trade dress.2

The district court's denial of Duraco's motion for a preliminary injunction was based on several grounds. First, it concluded that Duraco had failed to show a likelihood of success on the merits. The court disagreed with Duraco that the design of the Grecian Classics was distinctive. According to the court, "[a]ny consumer looking at the Duraco, Backyard [Products], and Cotswold planters would say that all three are Grecian urn planters." 822 F. Supp. 1202, 1207 (W.D.Pa.1993). It continued, remarking that "[t]o this observer, all of the urns introduced as exhibits, including plaintiff's, defendants' or others', looked about the same, as did those presented in drawings by Mr. Hart, the designer." Id.3

Applying a taxonomy of trademark distinctiveness consisting of generic, descriptive, suggestive, and arbitrary or fanciful marks, the court found that the Grecian Classics planters were descriptive at best: "The Duraco urn is clearly just that, an urn. When purchased or inspected, the consumer immediately knows the nature of the good before them [sic]: a plastic planter." Id. at 1209. The court further found that the planters' descriptive trade dress had not acquired secondary meaning in the market, as

[t]here was no evidence that consumers, whether K-Mart buyers or retail customers, perceived that the product emanates from a single source, or, for that matter, that the public identified the Duraco planter with the Duraco or Garden Scene name. The consumer does not appear to be moved in any degree to buy the Duraco planter because of its source. Buyers for the K-Mart stores seemed to be motivated largely by their profit margin.

Id. Alternatively, the court found the trade dress to be "generic." Id. at 1210.

The court also thought the design to be functional, "because [all the design elements are] necessary for classically styled urns, [and] in combination create the illusion of an alabaster chiseled urn." Id. Furthermore, the court found that there was no likelihood of consumer confusion, both because consumers are disinterested in the identity of the manufacturer and because there was no evidence of actual consumer confusion (neglecting that of Duraco's president, who is not truly a consumer) that Joy's product was Duraco's. Id.4

The district court had original jurisdiction pursuant to 28 U.S.C.A. Sec. 1338 (West 1993) (granting district courts original jurisdiction over trademark cases). Appellate jurisdiction is invoked pursuant to 28 U.S.C.A. Sec. 1292(a)(1) (West 1993) (granting courts of appeals jurisdiction over certain interlocutory orders, including denials of preliminary injunctions).

Duraco appeals the denial of its motion for a preliminary injunction in its action under section 43 of the Lanham Act, 15 U.S.C.A. Sec. 1125(a) (West Supp.1994).5 We explained the requisites for the issuance of a preliminary injunction as well as the nature of our review in such appeals in Merchant & Evans, Inc. v. Roosevelt Building Products Co., 963 F.2d 628 (3d Cir.1992):

In ruling on a motion for a preliminary injunction, the district court must consider: (1) the likelihood that the plaintiff will prevail on the merits at [the] final hearing; (2) the extent to which the plaintiff is being irreparably harmed by the conduct complained of; (3) the extent to which the defendant will suffer irreparable harm if the preliminary injunction is issued; and (4) the public interest. Opticians Ass'n v. Independent Opticians, 920 F.2d 187, 191-92 (3d Cir.1990). The injunction should issue only if the plaintiff produces evidence sufficient to convince the district court that all four factors favor preliminary relief. Id. at 192.

The decision whether to enter a preliminary injunction is committed to the sound discretion of the trial court, and will be reversed "only if the court abused its discretion, committed an obvious error in applying the law, or made a serious mistake in considering the proof." Loretangeli v. Critelli, 853 F.2d 186, 193 (3d Cir.1988). However, "[a]lthough terms of an injunction are normally reviewed for abuse of discretion, any determination that is a prerequisite to the issuance of an injunction ... is reviewed according to the standard applicable to that particular determination." John F. Harkins Co. v. Waldinger Corp., 796 F.2d 657, 658 (3d Cir.1986), cert. denied[,] 479 U.S. 1059, 107 S. Ct. 939 [93 L. Ed. 2d 989] (1987).

Id. at 632-33 (parallel citation omitted) (emphasis, and first and final alterations, supplied). Thus, we exercise plenary review over the district court's conclusions of law and its application of the law to the facts, see Marco v. Accent Publishing Co., 969 F.2d 1547, 1548 (3d Cir.1992), but review its findings of fact for clear error, see Oberti v. Board of Educ., 995 F.2d 1204, 1220 (3d Cir.1993), which occurs when we are "left with a definite and firm conviction that a mistake has been committed," Anderson v. Bessemer City, 470 U.S. 564, 573, 105 S. Ct. 1504, 1511, 84 L. Ed. 2d 518 (1985).

"Trade dress" originally referred to the packaging or displays associated with trademarked goods. In the leading case of Chevron Chemical Co. v. Voluntary Purchasing Groups, Inc., 659 F.2d 695, 700-02 (5th Cir.1981), for example, the Court of Appeals for the Fifth Circuit held that the defendant had unfairly competed with the plaintiff by utilizing the same design and colors as the plaintiff for packaging its lawn and garden products, even though the defendant prominently employed a different brand name.

That principle of trade dress law grounded in design protection has since been extended to the design of a product itself. "Section 43(a) of the Lanham Act proscribes not only trademark infringement in its narrow sense, but more generally creates a federal cause of action for unfair competition. In particular, Sec. 43(a) provides a cause of action for unprivileged imitation, including trade dress infringement [of unregistered trade dress]." American Greetings Corp. v. Dan-Dee Imports, Inc., 807 F.2d 1136, 1140 (3d Cir.1986) (citations omitted). That is, while it was once "well-settled that shapes of products in the public domain may be freely copied," id. at 1145 (citing cases),6 the same no longer holds true, as federal trademark law under the Lanham Act may protect "product configurations," see, e.g., Merchant & Evans, 963 F.2d at 633-34; cf. Aromatique, Inc. v. Gold Seal, Inc., 28 F.3d 863, 868 (8th Cir.1994) (per curiam) (plurality opinion by Morris Sheppard Arnold, J.) (noting that trade dress may now be registered on the Principal Register of the United States Patent and Trademark Office). In Merchant & Evans, we reaffirmed that trade dress protection extends beyond a product's packaging or labeling to include " 'the appearance of the [product] itself.' " Merchant & Evans, 963 F.2d at 633 (quoting American Greetings Corp., 807 F.2d at 1140).

The Lanham Act protection of product configurations extends to "the total image of a product, including features such as size, shape, color or color combinations, texture, graphics, or even particular sales techniques." Computer Care v. Service Sys. Enters., Inc., 982 F.2d 1063, 1067 (7th Cir.1992) (internal quotation marks omitted); accord International Jensen, Inc. v. Metrosound U.S.A., Inc., 4 F.3d 819, 822-23 (9th Cir.1993); Woodsmith Publishing Co. v. Meredith Corp., 904 F.2d 1244, 1247 (8th Cir.1990); Stormy Clime Ltd. v. ProGroup, Inc., 809 F.2d 971, 974 (2d Cir.1987); AmBrit, Inc. v. Kraft, Inc., 812 F.2d 1531, 1535 (11th Cir.1986); John H. Harland Co. v. Clarke Checks, Inc., 711 F.2d 966, 980 (11th Cir.1983). " 'Trade dress is a complex composite of features' and '[t]he law of unfair competition in respect to trade dress requires that all of the features be considered together, not separately.' " American Greetings Corp., 807 F.2d at 1141 (quoting SK & F, Co. v. Premo Pharmaceutical Labs., 481 F. Supp. 1184, 1187 (D.N.J.1979), aff'd, 625 F.2d 1055 (3d Cir.1980)). Thus, for example, a product configuration may be distinctive although no particular individual element or feature would be considered distinctive in isolation.

In this case we deal exclusively with trade dress said to inhere in the product itself, rather than trade dress alleged in a product's packaging. Because the legal doctrines in these two very different situations will substantially diverge in various incidents, we will employ the designation "product configuration" to refer to trade dress alleged in the product itself, whether in a specific feature or in some combination or arrangement of features, and to distinguish that type of trade dress from "product packaging." Cf. Two Pesos, Inc. v. Taco Cabana, Inc., --- U.S. ----, ----, 112 S. Ct. 2753, 2760, 120 L. Ed. 2d 615 (1992) ("[T]he protection of trademarks and trade dress under Sec. 43(a) serves the same statutory purpose of preventing deception and unfair competition. There is no persuasive reason to apply different analysis to the two.... It would be a different matter if there were textual basis in Sec. 43(a) for treating inherently distinctive verbal or symbolic trademarks differently from inherently distinctive trade dress.").

In Merchant & Evans, we held that in order to qualify for trade dress protection, the plaintiff must show:

(1) that the imitated feature is non-functional, (2) that the imitated feature has acquired a "secondary meaning," and (3) that consumers are likely to confuse the source of the plaintiff's product with that of the defendant's product.

Merchant & Evans, 963 F.2d at 633 (citing American Home Prods. Corp. v. Barr Labs., Inc., 834 F.2d 368, 370 (3d Cir.1987)); accord American Greetings Corp., 807 F.2d at 1141; Standard Terry Mills, Inc. v. Shen Mfg. Co., 803 F.2d 778, 780 (3d Cir.1986). The intervening decision in Two Pesos, Inc. v. Taco Cabana, Inc., --- U.S. ----, ----, 112 S. Ct. 2753, 2758, 120 L. Ed. 2d 615 (1992), has, however, overruled one aspect of that formulation, eliminating the need for the plaintiff to prove secondary meaning in order to receive protection for "inherently distinctive" trade dress. See Two Pesos, --- U.S. at ----, 112 S. Ct. at 2760. Hence a plaintiff proceeding under section 43(a) must now only prove that (i) the trade dress is distinctive, either because it is inherently distinctive or because it has acquired distinctiveness; (ii) the trade dress is nonfunctional; and (iii) the defendant's use of plaintiff's trade dress is likely to cause consumer confusion. See id.

As just stated, the Supreme Court in Two Pesos used "distinctive" in a dual sense, meaning either inherently distinctive or having acquired distinctiveness through secondary meaning. See RESTATEMENT (THIRD) OF UNFAIR COMPETITION Sec. 13 (Tent. Draft No. 2, Mar. 23, 1990),7 cited in Two Pesos, --- U.S. at ----, 112 S. Ct. at 2758. Because the Supreme Court in Two Pesos did not decide the question whether trade dress, and in particular trade dress in a product configuration, can actually ever be considered inherently distinctive--for purposes of that case, the Court assumed that the restaurant decor at issue was so--we must first embark on a journey to delineate when, if ever, product configurations should be deemed inherently distinctive.

Duraco argues that the design of its Grecian Classics planters is inherently distinctive because it is "suggestive." Duraco borrows the term "suggestive" from trademark law, as trademarks have long been classified according to whether they are generic, descriptive, suggestive, arbitrary, or fanciful. See Two Pesos, --- U.S. at ----, 112 S. Ct. at 2757 (citing Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 9 (2d Cir.1976) (Friendly, J.)). In trademark law, marks belonging to the latter three categories of the taxonomy--suggestive, arbitrary, or fanciful--are deemed inherently distinctive and are automatically entitled to protection. See id.; Fisons Horticulture, Inc. v. Vigoro Indus., Inc., 30 F.3d 466, 478 (3d Cir.1994); A.J. Canfield Co. v. Honickman, 808 F.2d 291, 296-97 (3d Cir.1986). Marks falling within the first category--generic marks--are never subject to trademark protection, because to tolerate their monopolization would preclude competitors from accurately and efficiently describing their products and hence unduly hobble them in competition. See Two Pesos, --- U.S. at ----, 112 S. Ct. at 2757; Kellogg Co. v. National Biscuit Co., 305 U.S. 111, 116-19, 59 S. Ct. 109, 112-14, 83 L. Ed. 73 (1938); Singer Mfg. Co. v. June Mfg. Co., 163 U.S. 169, 186, 16 S. Ct. 1002, 1008, 41 L. Ed. 118 (1896). The marks falling within the remaining category--descriptive--acquire distinctiveness only if they come to identify and distinguish the producer's goods, i.e., if they acquire secondary meaning. See 15 U.S.C.A. Secs. 1052(e)-(f), 1127 (West Supp.1994); Two Pesos, --- U.S. at ----, 112 S. Ct. at 2757. To have acquired secondary meaning, "in the minds of the public, the primary significance of a product feature or term [must be] to identify the source of the product rather than the product itself." Inwood Labs., Inc. v. Ives Labs., Inc., 456 U.S. 844, 851 n. 11, 102 S. Ct. 2182, 2187 n. 11, 72 L. Ed. 2d 606 (1982) (dicta).

Thus, Duraco seeks to latch onto the suggestive trademark category to avoid having to demonstrate secondary meaning. Some courts have nonchalantly applied the trademark generic/descriptive/suggestive/arbitrary/fanciful taxonomy in the product configuration context (though none of them has inquired whether it makes sense to do so), see, e.g., Braun, Inc. v. Dynamics Corp. of America, 975 F.2d 815, 825 n. 18 (Fed.Cir.1992) (design of a blender); cf. Paddington Corp. v. Attiki Importers and Distribs., Inc., 996 F.2d 577, 583 (2d Cir.1993) (product packaging); Computer Care, 982 F.2d at 1069 (same); AmBrit, Inc., 812 F.2d at 1537 (same), and both parties here assume that it provides the proper framework for our analysis, see Brief of Appellant at 22-32; Brief of Appellee at 22-30; Reply Brief of Appellant at 2-12. But we do not think it helpful or proper to transplant the categorical distinctiveness inquiry developed for trademarks to product configurations, where the alleged trade dress lies in the very product itself. See Martin P. Hoffman, Trade Dress/Product Simulation Overview, C913 ALI-ABA 219, 222 (1994) ("The [trademark distinctiveness] categories do not fit trade dress considerations very well....").

The difficulty is that, perhaps unlike product packaging, a product configuration differs fundamentally from a product's trademark, insofar as it is not a symbol according to which one can relate the signifier (the trademark, or perhaps the packaging) to the signified (the product). Being constitutive of the product itself and thus having no such dialectical relationship to the product, the product's configuration cannot be said to be "suggestive" or "descriptive" of the product, or "arbitrary" or "fanciful" in relation to it. See Jay Dratler, Jr., Trademark Protection for Industrial Design, 1988 U.ILL.L.REV. 887, 903 ("Unlike verbal marks, industrial designs do not describe anything; they 'just are.' "); Ralph S. Brown, Design Protection: An Overview, 34 UCLA L.REV. 1337, 1380 (1987) (same); Melissa R. Gleiberman, Note, From Fast Food to Fast Cars: Overbroad Protection of Product Trade Dress Under Section 43(a) of the Lanham Act, 45 STAN.L.REV. 2037, 2042-43 (1993) [hereinafter Gleiberman, Note, Overbroad Protection of Product Trade Dress ] (same); cf. WILLIAM H. BROWNE, A TREATISE ON THE LAW OF TRADE-MARKS Sec. 130, at 87 (1873) ("A trademark is nothing more nor less than one's commercial signature to his goods; and the mark and the goods bear the same relation to one another as do the positive and the negative forces of electricity to each other; and in their opposition they mutually uphold and sustain."). The very basis for the trademark taxonomy--the descriptive relationship between the mark and the product, along with the degree to which the mark describes the product--is unsuited for application to the product itself.

Moreover, insofar as consumer motivation to purchase a product will much more likely be predicated on an appreciation of a product's features than on an appreciation of a product's name, assuming no secondary meaning attached to either, one cannot automatically conclude from a product feature or configuration--as one can from a product's arbitrary name, for example--that, to a consumer, it functions primarily to denote the product's source. Cf. RESTATEMENT (THIRD) OF UNFAIR COMPETITION, Sec. 16 cmt. b (Tent. Draft No. 2, Mar. 23, 1990) ("[I]t is less common for consumers to recognize the design of a product as an indication of source."). As Judge Nies wrote, concurring in In re DC Comics, Inc., 689 F.2d 1042, 1050-51 (C.C.P.A.1982):

There are different considerations where one seeks protection of a product design itself, and I have found no precedent in decisions of this court, or others, which recognizes the protectability of any product design as a trademark for that product without proof of distinctiveness, that is, distinctiveness as an indication of origin, not simply that it is a distinctive design in the sense of being unusual. The semantics, in referring to a design as "distinctive," impedes clarity in analysis.... Descriptive designations are not presumed to function as indications of origin immediately upon first use, unlike arbitrary word marks or arbitrary logo designs, but rather must be used from some period of time before acquiring the status of a trademark.

Thus, a fanciful or arbitrary mark, having had no established meaning prior to its adoption as a trademark and serving no apparent purpose other than to identify (signify) the source, is legally presumed to achieve customer recognition and association immediately upon its adoption and use. In contrast, a product configuration can not generally give rise to a similar presumption, as consumers usually appreciate a product's configuration for its contribution to the inherent appeal of the product, not (in the absence of secondary meaning) its signifying function. If one felt compelled to apply the trademark taxonomy, one could at best say that a product configuration is descriptive of (because identical with) the product itself. This case illustrates the point rather clearly: whether or not the Grecian Classics are "suggestive" of a marble construction or anything else, we think it quite improbable that a consumer upon seeing Joy's plastic planter in a store would reasonably associate its specific configuration with a particular source, even if the consumer had repeatedly before seen a Duraco plastic planter.

Accordingly, for all the aforementioned considerations, we conclude that the trademark taxonomy, carefully and precisely crafted through a long succession of cases to accommodate the particularities of trademarks, does not fit the quite different considerations applicable to product configurations.8 See Hanig & Co. v. Fisher & Co., No. 92-C-1779, 1994 WL 97758, at * 4 (N.D.Ill. Mar. 24, 1994); Martin P. Hoffman, Trade Dress/Product Simulation Overview, C913 ALI-ABA at 222.

Although Two Pesos made extensive reference to the discrete trademark categories, that decision does not foreclose our refusal to embrace the trademark distinctiveness taxonomy in product configuration cases. In Two Pesos the Supreme Court specifically stated that the sole issue before it was whether secondary meaning must be proven for an inherently distinctive trade dress vel non, see --- U.S. at ---- & n. 6, 112 S. Ct. at 2757 & n. 6, and hence the applicability of the Abercrombie & Fitch classifications to trade dress was not at issue, see Paddington, 996 F.2d at 583. Moreover, Two Pesos dealt with a restaurant's decor, more akin to product packaging than product configuration.

Of course, the rationales supporting the trademark distinctiveness taxonomy may sometimes be fruitfully applied to trade dress when speaking of the product itself. For example, consider the trademark distinctiveness inquiry. What is "generic" in trademark law is a word with so few alternatives (perhaps none) for describing the good that to allow someone to monopolize the word would debilitate competitors. A descriptive trademark is one that leaves a larger but finite set of equivalent alternatives, and therefore still can be protected (because there are adequate alternatives for competitors) but only if it has acquired secondary meaning (so that it demonstrably functions as a source indicium). Finally, the suggestive, arbitrary, or fanciful mark is entitled to automatic protection, as there exists a vast universe of equivalent alternatives (or, in the case of a suggestive mark, at least a vast number of passable alternatives) to choose from, and the consumer will reasonably immediately identify the mark for what it is--a source indicium and no more.

As noted supra at 1439-40, a plaintiff alleging trade dress infringement must prove that the dress is nonfunctional.9 In trade dress law, the inquiry into functionality resembles the genericness inquiry in trademark law; the two doctrines share essentially the same underlying rationale, preserving competition. But cf. Dratler, Trademark Protection for Industrial Designs, 1988 U.ILL.L.REV. at 951-52 (arguing that the genericness doctrine and the functionality doctrine--in form substantially narrower than this Circuit has adopted, see supra at n. 9--serve different purposes). Thus, just as generic trademarks may be copied freely, functional trade dress may also be copied freely--because both are important for preserving effective competition. Similarly, just as descriptive trademarks, which are neither generic nor inherently distinctive, may be protected upon proof of secondary meaning, trade dress that is not functional (and thus leaves a satisfactory number of competitively viable alternatives, see Merchant & Evans, 963 F.2d at 634-35 & n. 4) but not inherently distinctive (and thus not a presumptive source indicium) may be protected, but only if secondary meaning is shown, see supra at 1439-40. Finally, just as inherently distinctive trademarks are protected because presumptively they identify the product's source, inherently distinctive trade dress is protected because presumptively it primarily identifies the product's source.

C. Distinctiveness in the Law of Unfair Competition

Having rejected the transplantation of the trademark distinctiveness categories to product configurations, the next question we must address is whether a product configuration can ever be inherently distinctive. Before the Supreme Court's decisions in Sears, Roebuck & Co. v. Stiffel Co., 376 U.S. 225, 84 S. Ct. 784, 11 L. Ed. 2d 661 (1964), and Compco Corp. v. Day-Brite Lighting, Inc., 376 U.S. 234, 84 S. Ct. 779, 11 L. Ed. 2d 669 (1964), unfair competition law was essentially state law. Sears and Compco created a federal unfair competition law. See Two Pesos, --- U.S. at ----, 112 S. Ct. at 2763 (Stevens, J., concurring in the judgment) (" 'Section 43(a) ... has been widely interpreted to create, in essence, a federal law of unfair competition.' ") (quoting The United States Trademark Ass'n Trademark Rev. Comm'n Report and Recommendations to USTA President and Board of Directors, 77 TRADEMARK RPTR. 375, 426 (1987)); American Greetings Corp., 807 F.2d at 1140 ("Section 43(a) ... creates a federal cause of action for unfair competition."); Kohler Co. v. Moen, Inc., 12 F.3d 632, 646 (7th Cir.1993) (Cudahy, J., dissenting) ("[T]he Lanham Act (comprising the federal law of trademarks and unfair competition) essentially federalizes the common law of trademarks and unfair competition."); id. at 647; cf. Kohler Co., 12 F.3d at 640 n. 10 (recognizing relationship between Lanham Act and state common law of unfair competition but emphasizing that they are not identical). Congress approved that development in the 1988 amendments to the Lanham Act. See Two Pesos, --- U.S. at ----, 112 S. Ct. at 2765 (Stevens, J., concurring in the judgment) ("Congress codified the judicial interpretation of Sec. 43(a), giving its imprimatur to a growing body of case law...."); Kohler Co., 12 F.3d at 636 ("[The 1988 Lanham Act amendments] as a codification of prior case law ... validate the uniform preamendment interpretation of Sec. 45 [of] the Act."); S.REP. NO. 515, 100th Cong., 2d Sess. 40 (1988), reprinted in 1988 U.S.C.C.A.N. 5577, 5603. Thus, it will be instructive to commence our inquiry with a survey of the common law tort of unfair competition.

Under state unfair competition law, a product configuration could obtain protection from copying only if it first had acquired secondary meaning. See, e.g., Gum, Inc. v. Gumakers of America, Inc., 136 F.2d 957, 958 (3d Cir.1943); American Fork & Hoe Co. v. Stampit Corp., 125 F.2d 472, 474 (6th Cir.1942); Lewis v. Vendome Bags, Inc., 108 F.2d 16, 18 (2d Cir.1939); Sinko v. Snow-Craggs Corp., 105 F.2d 450, 452 (7th Cir.1939); Crescent Tool Co. v. Kilborn & Bishop Co., 247 F. 299, 300 (2d Cir.1917); Rathbone, Sard & Co. v. Champion Steel Range Co., 189 F. 26, 30-32 (6th Cir.1911); 1 HARRY D. NIMS, THE LAW OF UNFAIR COMPETITION AND TRADE-MARKS Sec. 134(a), at 370, 378-79 (4th ed. 1947) [hereinafter NIMS, LAW OF UNFAIR COMPETITION]; see also Two Pesos, --- U.S. at ----, 112 S. Ct. at 2762 (Stevens, J., concurring in the judgment); Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141, 158, 109 S. Ct. 971, 981, 103 L. Ed. 2d 118 (1989); 1 J. THOMAS MCCARTHY, TRADEMARKS AND UNFAIR COMPETITION Sec. 7:23, at 227, Sec. 8:2, at 285 (2d ed. 1984). The concept of an inherently distinctive product configuration was, in other words, alien to the common law.

Secondary meaning in the product configuration context was then, as it is now, defined to "attach[ ] to a given shape or form of article when that form is associated in the minds of prospective customers with the source from which the article came to such an extent that demand for the particular article depends upon the business reputation or standing of its maker." American Fork & Hoe, 125 F.2d at 475; see Kellogg Co., 305 U.S. at 118, 120, 59 S. Ct. at 113-14 (explaining that secondary meaning has not been established when the form of the article, in the minds of the public, is primarily associated with the article rather than a particular producer); supra at typescript 15-16 (quoting Inwood Labs.). The Court of Appeals for the Seventh Circuit succinctly expounded the persuasive rationale for this doctrine in Sinko, which involved the defendant's copying of plaintiff's successful knobs for automobile steering wheel spinners:

Sinko created a desire on the part of the public for one of two things, either for knobs made by Sinko, above all other knob makers, or for knobs made in a particular manner regardless of who made them. If it is the first situation, the law of unfair competition gives Sinko the right to monopolize or to exclude other makers from copying the product. If it is the latter situation, Sinko receives no such right to monopolize, even though he might have been the first one to make the article in the particularly desirable manner.

Sinko, 105 F.2d at 453; accord 1 NIMS, LAW OF UNFAIR COMPETITION Sec. 134, at 373 (citing Diamond Expansion Bolt Co. v. U.S. Expansion Bolt Co., 177 A.D. 554, 164 N.Y.S. 433 (1917)). Similarly, then-District Judge Learned Hand, sitting by designation, wrote for the Second Circuit:

The cases of so-called "nonfunctional" unfair competition ... are only instances of the doctrine of "secondary meaning." All of them presuppose that the appearance of the article, like its descriptive title in true cases of "secondary" meaning, has become associated in the public mind with the first comer as manufacturer or source, and, if a second comer imitates the article exactly, that the public will believe his goods have come from the first, and will buy, in part, at least, because of that deception. Therefore it is apparent that it is an absolute condition to any relief whatever that the plaintiff in such cases show that the appearance of his wares has in fact come to mean that some particular person--the plaintiff may not be individually known--makes them, and that the public cares who does make them, and not merely for their appearance and structure. It will not be enough only to show how pleasing they are, because all the features of beauty or utility which commend them to the public are by hypothesis already in the public domain. The defendant has as much right to copy the "nonfunctional" features of the article as any others, so long as they have not become associated with the plaintiff as manufacturer or source. The critical question of fact at the outset always is whether the public is moved in any degree to buy the article because of its source and what are the features by which it distinguishes that source.

Crescent Tool Co., 247 F. at 300 (emphasis supplied).

Finally, Chief Justice Holmes, then sitting on the Supreme Judicial Court of Massachusetts, also explained the irreproachable rationale sustaining the prerequisite of secondary meaning before affording protection to product configurations:

In the absence of a patent the freedom of manufacture cannot be cut down under the name of preventing unfair competition. All that can be asked is that precautions shall be taken, so far as are consistent with the defendant's fundamental right to make and sell what it chooses, to prevent ... deception....

It is true that a defendant's freedom of action with regard to some subsidiary matter of ornament or label may be restrained, although a right of the same nature with its freedom to determine the shape of the articles which it sells.... [T]he instrument sold is made as it is [by defendant], partly at least, because of a supposed or established desire of the public for instruments in that form. The defendant has the right to get the benefit of that desire even if created by the plaintiff. The only thing it has not the right to steal is the good will attaching to the plaintiff's personality, the benefit of the public's desire to have goods made by the plaintiff.... [T]he plaintiff's right can be protected sufficiently by requiring the defendant's [products] to be clearly marked so as to indicate unmistakably that they are the defendant's and not the plaintiff's goods.... To go further is to save the plaintiff from a competition from which it has no right to be exempt.

Flagg Mfg. Co. v. Holway, 178 Mass. 83, 59 N.E. 667, 667 (1901) (citations omitted).

Thus, traditional unfair competition law would not mandate a copier to take positive steps to avoid confusion unless "the existence of secondary meaning ... plainly appear[ed]." American Fork & Hoe, 125 F.2d at 475. If a product feature had obtained secondary meaning, but was functional, all a defendant had to do, to avoid competing unfairly, was to "use reasonable care to inform the public of the source of its product." Gum, Inc., 136 F.2d at 960 (citing Kellogg Co., 305 U.S. at 120, 59 S.Ct. at 114); see Vaughan Novelty Mfg. Co. v. G.G. Greene Mfg. Corp., 202 F.2d 172, 176 & n. 11 (3d Cir.) (citing RESTATEMENT OF TORTS Sec. 741, cmt. j (1938)), cert. denied, 346 U.S. 820, 74 S. Ct. 34, 98 L. Ed. 346 (1953); American Fork & Hoe, 125 F.2d at 475; Crescent Tool Co., 247 F. at 301; 1 NIMS, LAW OF UNFAIR COMPETITION Sec. 134, at 371; cf. J.C. Penney Co. v. H.D. Lee Mercantile Co., 120 F.2d 949, 955-56 (8th Cir.1941). This latter doctrine survives intact today. See American Greetings Corp., 807 F.2d at 1141 ("When a feature or combination of features is found to be functional, it may be copied and the imitator may not be enjoined from using it.... Nevertheless, if the functional feature or combination is also found to have acquired secondary meaning, the imitator may be required to take reasonable steps to minimize the risk of source confusion.") (citations omitted); id. at 1144-45 & n. 4. A defendant could be completely barred from copying a product configuration only if the configuration both was nonfunctional and had acquired secondary meaning.

In short, the common law did not find any "unfairness," as concerns the law, in someone's copying a design--even if it was originally produced through great expenditures of labor, effort, talent, and capital--if the design was unprotected by patent or copyright. See, e.g., Zangerle & Peterson Co. v. Venice Furniture Novelty Mfg. Co., 133 F.2d 266, 269 (7th Cir.1943). What the courts of equity condemned was not bare-knuckled competition, but fraud and deceit, which are worked when one "palms off" one's goods as those of another, see, e.g., Zangerle & Peterson, 133 F.2d at 269-70; J.C. Penney, 120 F.2d at 953-54; Lewis, 108 F.2d at 18; Sinko, 105 F.2d at 452; Rathbone, Sard & Co., 189 F. at 30-31; EDWARD S. ROGERS, GOOD WILL, TRADE-MARKS, AND UNFAIR TRADING 212 (1914); cf. 1 NIMS, LAW OF UNFAIR COMPETITION Sec. 134, at 374 ("The fact that the defendant may have deliberately copied the appearance of the plaintiff's goods and that in doing so may have gained an advantage, is not enough. The imitating must have been done with the expectation of obtaining some unfair advantage."); see also Two Pesos, --- U.S. at ---- & n. 5, 112 S. Ct. at 2762 & n. 5 (Stevens, J., concurring in the judgment); Bonito Boats, 489 U.S. at 157, 109 S.Ct. at 981; that is, when one deceives the consuming public and misappropriates--trades upon--the good will of another. Exploiting the "goodwill of the article," Kellogg Co., 305 U.S. at 121, 59 S.Ct. at 115--the attractive features, of whatever nature, that the product holds for consumers--is robust competition; only deceiving consumers, or exploiting the good will of another producer, is unfair competition.

Besides having no foothold in the common law, recognizing the existence of inherently distinctive product configurations is arguably inconsistent with precedent in this Circuit. As we have mentioned, in Merchant & Evans we held that a feature of the product itself qualifies for trade dress protection only if the plaintiff proves that the imitated feature is nonfunctional, that it has acquired secondary meaning, and that consumers are likely to confuse the source of the plaintiff's products with that of the defendant's. Merchant & Evans, 963 F.2d at 633. This might be read as an implied holding that trade dress in the product itself can never be inherently distinctive. But the panel did not advance this proposition in Merchant & Evans, and, especially given the Supreme Court's intimations in the intervening Two Pesos decision, we do not read Merchant & Evans in that fashion.

As already stated, Two Pesos poses a problem with adhering to Merchant & Evans insofar as the Supreme Court's decision eliminated the secondary meaning prong for inherently distinctive trade dress. Of course, Two Pesos only answered the question "whether trade dress which is inherently distinctive is protectable under Sec. 43(a) without a showing that it has acquired secondary meaning," --- U.S. at ----, 112 S. Ct. at 2757 (emphasis supplied); it did not define what makes trade dress inherently distinctive and, more importantly for our purposes, did not decide whether a product configuration could ever be inherently distinctive, cf. Kohler Co., 12 F.3d at 641 n. 11 (dicta) ("[A] product's shape is never inherently distinctive."). And Two Pesos, which dealt with the decor of a Mexican restaurant, is a product packaging, not a product configuration, case. See --- U.S. at ---- & n. 1, 112 S. Ct. at 2755 & n. 1. See also RESTATEMENT (THIRD) OF UNFAIR COMPETITION Sec. 16 cmt. b (Tent. Draft No. 2, Mar. 23, 1990) ("Product designs are ... not considered inherently distinctive; such designs are protectable only upon proof of secondary meaning."); 1 MCCARTHY, TRADEMARKS AND UNFAIR COMPETITION Sec. 7.31 n. 1 (citing conflicting sources).

Nevertheless, we do read Two Pesos as giving an imprimatur to finding trade dress in a product configuration to be inherently distinctive under certain narrow circumstances. Cf. id. at ----, 112 S.Ct. at 2759 (not distinguishing among "a verbal or symbolic mark or the features of a product design"); id. at ----, 112 S. Ct. at 2760 (discussing protectability of "design[s]" and "shape[s]"). The Supreme Court provided two strong competition-based rationales why at least some trade dress should be deemed inherently distinctive.

First, the Court noted that protection of trade dress along with trademarks could further the Lanham Act's purpose to " 'secure to the owner of the mark [or dress] the goodwill of his business and to protect the ability of consumers to distinguish among competing producers.' " Two Pesos, --- U.S. at ----, 112 S. Ct. at 2760 (quoting Park 'N Fly, Inc. v. Dollar Park and Fly, Inc., 469 U.S. 189, 198, 105 S. Ct. 658, 663, 83 L. Ed. 2d 582 (1985)). "By making more difficult the identification of a producer with its product," the Court explained, "a secondary meaning requirement for a non-descriptive trade dress would hinder improving or maintaining the producer's competitive position." Id. We think that such an overly rigorous uniform requirement for product configurations would have a similar deleterious effect.

Second, the Court reasoned that always requiring secondary meaning would not protect the developer of a "fanciful" or "arbitrary" trade dress from "theft" while the developer tries to acquire secondary meaning for its trade dress. See id. at ----, 112 S. Ct. at 2758-59; cf. Keene Corp. v. Paraflex Indus., Inc., 653 F.2d 822, 825 (3d Cir.1981). In particular, the Court was concerned about the "anticompetitive effects [of] particular burdens on the start-up of small companies." Two Pesos, --- U.S. at ----, 112 S. Ct. at 2761. Always requiring secondary meaning would burden fledgling companies seeking to expand into new markets because established competitors could "appropriate the originator's [inherently distinctive nonfunctional trade] dress in other markets and ... deter the originator from expanding into and competing in these areas." Id. This rationale also would seem to apply whether the trade dress is alleged in a product packaging or a product configuration.

Of course, it is not the purpose of unfair competition law, under the guise of either consumer protection or the protection of business good will, to implement a policy of encouraging innovative designs by protecting them once designed.10 See Gleiberman, Note, Overbroad Protection of Product Trade Dress, 45 STAN.L.REV. at 2056-57; cf. Dratler, Trademark Protection for Industrial Designs, 1988 U.ILL.L.REV. at 909 ("[F]ostering innovation is only a minor and incidental purpose of trademark protection."). Those issues are the province of copyright and patent laws. Moreover, design protection laws (which have repeatedly been introduced in Congress during virtually every session since 1917) have not once been enacted. See In re Nalbandian, 661 F.2d 1214, 1218-19 & n. 1 (C.C.P.A.1981) (Rich, J., concurring) (recounting the history of efforts to pass design protection legislation); Esquire, Inc. v. Ringer, 591 F.2d 796, 800 n. 12 (D.C.Cir.1978) (observing that since 1914 none of the approximately 70 design protection bills introduced in Congress had passed); Brown, Design Protection: An Overview, 34 UCLA L.REV. at 1395 ("Beginning in 1957, a [design protection] bill has been introduced in probably every Congress...."); Dratler, Trademark Protection for Industrial Designs, 1988 U.ILL.L.REV. at 888 & n. 4, 904 & n. 94 (stating that the history of the effort to have Congress enact industrial design legislation takes up 160 pages in a Copyright Office bibliography, and that between 1914 and 1945 at least 45 such bills were introduced in Congress); A. Samuel Oddi, The Functions of "Functionality" in Trademark Law, 22 HOUS.L.REV. 925, 951 & n. 147 (1985); Gleiberman, Note, Overbroad Protection of Product Trade Dress, 45 STAN.L.REV. at 2070 & nn. 251-52; cf. Bonito Boats, 489 U.S. at 167-68, 109 S.Ct. at 986 ("It is for Congress to determine if the present system of design and utility patents is ineffectual in promoting the useful arts in the context of industrial design.").

Under the laws that are on the books, Congress has repeatedly chosen not to protect designs unless they meet certain strict requirements, catalogued in Merchant & Evans, 963 F.2d at 638-40; see also Dratler, Trademark Protection for Industrial Designs, 1988 U.ILL.L.REV. at 923-35; Gleiberman, Note, Overbroad Protection of Product Trade Dress, 45 STAN.L.REV. at 2055-65, and then Congress has only given them protection of circumscribed duration, cf. U.S. CONST. art. I, Sec. 8, cl. 8 (authorizing Congress to grant patents of limited duration). We believe that courts should exercise restraint so as not to undermine Congress's repeated determinations not to afford virtually perpetual protection to product configurations with an expansive construction of section 43(a). What Congress has, for the great span of this century, been unwilling to do, see supra at n. 10, should not be effected by the judiciary.

Thus Duraco's suggestion, based on its reading of Paddington, 996 F.2d at 582-83, that the capacity of the product's configuration to distinguish the plaintiff's goods from others suffices to establish inherent distinctiveness, is grossly overinclusive. It is also circular: clearly any perceptible product feature or combination or arrangement of features can distinguish goods, and perhaps is likely to do so if, as a rule, nobody else were allowed to copy it.11 That is, provided that no one besides the originator is allowed to use a particular feature, it would be difficult to conjure up any perceptible feature that users can train upon that is not capable of distinguishing the originator's goods from those of others. For example, even the basic design of a light bulb is "capable of identifying a particular source of the product," Paddington, 996 F.2d at 582-83, assuming that only one manufacturer produces the basic design, a fact which would be assured, of course, if the design were protected against copying.

Duraco's proposal to treat any product feature or configuration as inherently distinctive if it were merely capable of identifying the source of the product would therefore eviscerate the requirement for showing secondary meaning. Cf. Turtle Wax, Inc. v. First Brands Corp., 781 F. Supp. 1314, 1321 (N.D.Ill.1991) ("Presumably, it could be said about the trade dress of any new product that no competitive product combines precisely the same elements in its trade dress.... However, that fact alone does not make the product's trade dress inherently distinctive. Any other rule essentially would require a finding of inherent distinctiveness whenever a new product enters the market."). This is so even though the design is not at all deserving of such protection because, leaving aside the question of functionality, a consumer seeing another good incorporating the same design feature or features would not reasonably believe the first manufacturer to be its source.

Moreover, contrary to Duraco's understanding, Paddington did not read Two Pesos as establishing the "capable of identifying" standard for inherent distinctiveness. In the Two Pesos passages cited in Paddington, the Supreme Court was simply establishing that inherently distinctive trade dress need not acquire secondary meaning before obtaining protection under the Lanham Act. Thus, the Supreme Court stated that no secondary meaning would be required for "inherently distinctive trade dress ... capable of identifying a producer's product." --- U.S. at ----, 112 S. Ct. at 2760 (emphases supplied). That statement explains what is necessary, but not what is sufficient, to make trade dress inherently distinctive. Paddington, alert to this qualification, simply held that a feature's capacity to distinguish the goods is a prerequisite to trademark protection, not a sufficient condition. See 996 F.2d at 582-83. Indeed, the Court of Appeals for the Second Circuit in Paddington--a product packaging case--went further and distinguished between merely descriptive trade dress, for which the plaintiff had to prove secondary meaning, and suggestive, fanciful, or arbitrary trade dress, for which no secondary meaning was required. See 996 F.2d at 583; see also Nancy D. Chapman, Trade Dress Protection in the United States After the Supreme Court Decision in Two Pesos, 387 PLI/PAT 7, 13-16, 40-42 (1994) [hereinafter Chapman, Trade Dress Protection ] (criticizing the "capable of identifying" standard).

In any event, the analysis appropriate for a product's packaging, at issue in Paddington, is not necessarily appropriate for a product's configuration. Product packaging designs, like trademarks, often share membership in a practically inexhaustible set of distinct but approximately equivalent variations, and an exclusive right to a particular overall presentation generally does not substantially hinder competition in the packaged good, the item in which a consumer has a basic interest. A product configuration, contrariwise, commonly has finite competitive variations that, on the whole, are equally acceptable to consumers. See RESTATEMENT (THIRD) OF UNFAIR COMPETITION Sec. 16 cmt. b (Tent. Draft No. 2, March 23, 1990) (contrasting product configurations and product packagings based on the availability of alternatives). Moreover, because of consumers' common abundant experience with similar goods being sold in differing packaging, a consumer is substantially more likely to trust a product's packaging, rather than its configuration, as an indicium of source. Accordingly, we reject the misreading of Paddington advanced by Duraco.

It is not ipso facto "unfair competition," we believe, for one boldly to copy a competitor's product; it is only "unfair competition" to trade off another's good will and in the process dupe consumers into mistaking one's products for another's. A proper approach to inherent distinctiveness must distinguish between nonfunctional but desirable designs--which, absent secondary meaning, unfair competition law has no interest in precluding others from copying--and nonfunctional designs representing to consumers the source of the goods--which unfair competition law does and should forbid others from copying.

3. The Standard for Inherent Distinctiveness of Trade Dress in Product Configurations

Synthesizing the principles explored in the preceding sections, we think that there is a proper set of circumstances for treating a product configuration as inherently distinctive. These circumstances are characterized by a high probability that a product configuration serves a virtually exclusively identifying function for consumers--where the concerns over "theft" of an identifying feature or combination or arrangement of features and the cost to an enterprise of gaining and proving secondary meaning outweigh concerns over inhibiting competition, and where consumers are especially likely to perceive a connection between the product's configuration and its source. In particular, we think that, to be inherently distinctive, a product configuration--comprising a product feature or some particular combination or arrangement of product features--for which Lanham Act protection is sought must 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.

For a product configuration to have the capacity to distinguish goods in a consumer's mind--the first prerequisite for inherent distinctiveness--it must be unusual and memorable. It must partake of a unique, individualized appearance, so that a consumer informed of all the options available in the market could reasonably rely on it to identify a source. See Computer Care, 982 F.2d at 1069; Hoffman, Trade Dress/Product Simulation Overview, C913 ALI-ABA at 222 (asserting that trade dress is inherently distinctive only if "so unique, ... in a particular market, that one can assume, without proof, that it will automatically be perceived by customer[s] as an indicia [sic] of origin--a trademark") (omission in original) (citing 1 MCCARTHY, TRADEMARKS AND UNFAIR COMPETITION and AmBrit, Inc. v. Kraft, Inc., 805 F.2d 974, opinion superseded by 812 F.2d 1531 (11th Cir.1986)). "Manifestly, if the plaintiff's trade dress is not sufficiently distinctive to allow consumers to identify the [source] from the trade dress, then the dress does not inherently serve as an indication of origin...." AmBrit, Inc., 812 F.2d at 1536, quoted in NExxUS Prods. Co. v. Gentle Concepts, Inc., 28 U.S.P.Q.2d 1257, 1266, 1993 WL 496824 (M.D.Fla.1993).

Moreover, unless the trade dress is memorable--that is, striking or unusual in appearance, or prominently displayed on the product packaging, or otherwise somehow apt to be impressed upon the minds of consumers, so that it is likely to be actually and distinctly remembered--it cannot serve as a designator of origin. See Stuart Hall Co. v. Ampad Corp., 31 U.S.P.Q.2d 1468, 1470, 1994 WL 228939 (W.D.Mo.1994); id. at 1471 ("The trade dress must be remembered before it can be confusing."); cf. AmBrit, Inc., 812 F.2d at 1536 (setting forth criterion to measure distinctiveness); Computer Care, 982 F.2d at 1069 (holding that sales brochures are inherently distinctive if consumers can identify the product from the trade dress and the trade dress is arbitrary or suggestive). Thus, for example, designs customary in the industry can not be inherently distinctive (nor for that matter can they acquire secondary meaning). See Paddington, 996 F.2d at 583-84; Hanig & Co., 1994 WL 97758, at * 6-* 7; 1 NIMS, LAW OF UNFAIR COMPETITION Sec. 134a, at 379 (citing Rathbone, Sard & Co., 189 F. 26); cf. Aromatique, 28 F.3d at 869-70 (per curiam) (plurality opinion by Morris Sheppard Arnold, J.) (holding that the plaintiff's pillow-shaped cellophane packages were not inherently distinctive, because commonplace).

But the uniqueness of a product configuration is not enough by itself to make the configuration inherently distinctive. To be inherently distinctive, a product configuration must also be conceptually separable from the product, so that a consumer will recognize its symbolic (signifying) character. This requirement ensures that consumers unaware of any association of the product with a manufacturer (i.e., where a configuration has no secondary meaning) will not become confused about whether a particular configuration may be trusted as an indicium of origin. To be conceptually separable, the product configuration must be recognizable by the consumer "as an indicium of source, rather than a decorative symbol or pattern...." Stuart Hall, 31 U.S.P.Q.2d at 1471, 1994 WL 228939 at * 4 (internal quotation marks omitted); see Hoffman, Trade Dress/Product Simulation Overview, C913 ALI-ABA at 223 (contending that trade dress is inherently distinctive only if "one can assume, without proof, that it will automatically be perceived by customers as an indicia [sic] of origin"); cf. 1 MCCARTHY, TRADEMARKS AND UNFAIR COMPETITION Sec. 7.26, at 248 (suggesting that features perceived as "merely ornamental" cannot serve as trademarks).

As with trademarks, an inherently protectable product configuration must, at least conceptually, be "something other than, and separate from, the merchandise." Davis v. Davis, 27 F. 490, 492 (C.C.Mass.1886). That is, the configuration for which protection is sought must not appear to the consumer as a mere component, or the essence, of the product gestalt, but rather must appear as something attached (in a conceptual sense) to function in actuality as a source designator--it must appear to the consumer to act as an independent signifier of origin rather than as a component of the good. See, e.g., In re General Tire & Rubber Co., 404 F.2d 1396, 1398-99 (C.C.P.A.1969) (affirming denial of injunction because consumers would probably think of product feature as ornamentation rather than as indicium of source).

Third, to be inherently distinctive, it must be likely that the product configuration will primarily serve as a designator of the source of the product. See Chapman, Trade Dress Protection, 387 PLI/PAT at 32 (" 'Unique' is not by itself equivalent to inherently distinct, but a unique design may be so if it also functions as a source indicator."); supra at typescrpt 29-31 (rejecting Duraco's reading of Paddington ). If the configuration itself, separate from the product, is likely to serve some substantial purpose other than as a designation of origin--that is, besides to set it apart from other sources' products in consumers' minds--then it cannot be inherently distinctive, but must acquire secondary meaning before becoming entitled to protection against copying. In this regard, a source's intent in adopting the particular configuration is highly probative. See Stuart Hall, 31 U.S.P.Q.2d at 1470 & n. 2, 1994 WL 228939 at * 3-* 4; RESTATEMENT OF TORTS Sec. 719 cmt. a (1938) ("[D]esigns intended solely as ornamentations are not trade-marks...."); Joan L. Dillon, Two Pesos: More Interesting for What It Does Not Decide, 83 TRADEMARK RPTR. 77, 85 (1993) (defining protectable trade dress as "a combination of elements selected to identify origin, rather than to serve as mere decor"); 1 NIMS, LAW OF UNFAIR COMPETITION, at 370 ("Peculiar and arbitrary features in the form, structure, arrangement of parts, and general appearance of an article are often devised, in large part to distinguish it--to give it individuality.").

The inquiry here does not duplicate that employed for secondary meaning; instead of focusing on consumers' actually acquired mental associations, the inquiry focuses on whether a consumer would likely perceive the feature or combination or arrangement of features as something that renders the product intrinsically more desirable regardless of the source of the product, or primarily as a signifier of the product's source. Protecting a product configuration without a showing of secondary meaning because the configuration is pleasant rather than because it identifies the source of the product would risk seriously transgressing the protective zones mapped by the patent laws.12 This is not to say that a configuration must be un desirable to be eligible as an inherently distinctive product configuration; we refer here only to the source-designating function of the configuration.

Thus, if the feature as to which Lanham Act protection is sought is likely to be notably desirable to consumers for some reason other than its function as a source designator, it cannot be considered inherently distinctive. Like a descriptive trademark, such a nonfunctional product configuration would be protectable only if the source-designating function of the configuration is factually demonstrated with proof of secondary meaning. So if the consumer is likely to be motivated, in some more than incidental part, to buy a product because of a particular combination or arrangement of features, other than because that configuration signifies a source of the product, the penumbra of the patent laws--granting others a right to copy what has been donated to the public domain--will deny protection unless secondary meaning is first shown.

The primarily source-designating inquiry is different from the functionality inquiry in trade dress law, although it is motivated by similar concerns. Both are intended to protect competition--and hence consumers--by restricting the types of product features that may be insulated from copying. Nevertheless, the primarily source-designating inquiry is different from the functionality inquiry, for it limits inherently distinctive product configurations not to those that are not important for competitors to be able to copy, but rather to those whose primary significance is as an intrinsic indicator of the product's source.

We acknowledge that, to a large extent, how courts resolve the inherent distinctiveness inquiry could, theoretically at least, cause a snowballing effect. If product configurations are easily protected, consumers might learn to rely on configurations as source designators; if protection is rare, consumers will disregard product configurations as source designators, and no confusion will result. But partial protection, if not carefully circumscribed, may eventually cause even greater consumer confusion, as consumers will face difficulties determining what features are legitimate source designators (because inherently distinctive) and which are not. The narrow test that we adopt encourages consumers to rely on a product's configuration as a source designator only when it rather plainly serves an identifying function.

We believe that the aforementioned requirements for inherent distinctiveness, while nascent and in need of case law development, allow a source genuinely intent upon using a particular feature or configuration of its product to signify itself as the source to do so--at no meaningful cost to free competition--without having to prove consumer association (secondary meaning). We turn to the application of this standard to the facts of this case.

Because the district court applied the Abercrombie & Fitch trademark taxonomy (generic, descriptive, suggestive, arbitrary, or fanciful) instead of the above standard, see 822 F. Supp. at 1208-10, we cannot let its finding of non-distinctiveness stand. While this would normally result in a remand, we need not remand if, with the evidence viewed in the light most favorable to it, Duraco has failed to demonstrate a likelihood of success on the merits. See In re Molded Acoustical Prods., Inc. (Fiber Lite Corp. v. Molded Acoustical Prods., Inc.), 18 F.3d 217, 222 & n. 7 (3d Cir.1994). We conclude that this is the case, for Duraco's Grecian Classics planters plainly are not inherently distinctive, and it has not shown the acquired distinctiveness (secondary meaning) that it would otherwise need to establish in order for its product configuration to be protectable.

We do not gainsay that the particular configuration of Duraco's urn may be memorable and unusual. In this regard, we disagree with the district court's conclusion that "all of the urns introduced as exhibits ... looked about the same, as did those presented in drawings," 822 F. Supp. at 1207. Instead, after viewing most of the plastic planter designs introduced in the district court, including the physical exhibits and drawings, we believe that some definitely have a look that consumers could not reasonably confuse with the Grecian Classics, Backyard Products, or Cotswold planters. While mere distinguishability without more is, as we have explained, insufficient to establish that a product configuration is memorable and unusual in the sense that it could be an inherently distinctive indicium of source, we cannot say with certainty that Duraco's planters were not memorable and unusual at the time that Joy introduced the Ultimate Urns into interstate commerce.

Our review of course is deferential. But we need not run out the string in our review of the district court's fact finding in this respect for, at all events, we are satisfied that the Grecian Classics' design rather clearly falls outside the inherently distinctive category.

First, it is beyond peradventure that Duraco's alleged protectable product configuration is not conceptually separable from its plastic planters. Unlike an arbitrary carving etched into the back of a chair to identify the source, for example, the features of the Grecian Classics for which Duraco seeks protection are designed to achieve the goal of having the planter appear as if constructed of marble or stone, and thus constitute part and parcel of the overall product.13

Second, consumers are likely to appreciate the Grecian Classics' design, as the district court found to be the case, primarily as an inherently attractive aspect of the product, not as a source-indicator. Abundant and uncontroverted testimony established that Duraco adopted its features to create the appearance of a marble, stone, or cement Grecian or classical-style urn, not to identify itself as the source. Indeed, Duraco itself admits that consumers are largely motivated to purchase Grecian Classics because of the aesthetic advantages of the precise configuration for which it seeks protection. Moreover, Joy imitated the Grecian Classics' design because of its aesthetic appeal to consumers.

In sum, two of the requirements for inherent distinctiveness not having been satisfied, we easily conclude that Duraco's planter is not inherently distinctive.

B. Acquired Distinctiveness (Secondary Meaning)

The absence of inherent distinctiveness means that to establish a likelihood of success on the merits of its trade dress infringement claim, Duraco must prove secondary meaning, which the district court found it had not done. See id. at 1209. Although the point is without difficulty, we will briefly review it for clear error.

If a product's configuration is not inherently distinctive, the plaintiff must prove acquired distinctiveness via a showing of secondary meaning. Factors relevant to a finding of secondary meaning in a product configuration include: (1) plaintiff's advertising expenditures, measured primarily with regard to those advertisements which highlight the supposedly distinctive, identifying feature, see First Brands Corp. v. Fred Meyer, Inc., 809 F.2d 1378, 1383 (9th Cir.1987); (2) consumer surveys linking the distinctive product configuration to a particular, single source (although the identity of the source need not be known); and (3) length and exclusivity of use. Consumer surveys and testimony are probably the only direct evidence of secondary meaning; the other sources are circumstantial, though the plaintiff may rely solely on them. See Aromatique, 28 F.3d at 871 (per curiam) (plurality opinion by Morris Sheppard Arnold, J.) (secondary meaning for trade dress); Woodsmith Publishing Co., 904 F.2d at 1249 (same); International Kennel Club, Inc. v. Mighty Star, Inc., 846 F.2d 1079, 1086 (7th Cir.1988) (secondary meaning for trademark).

Sales success by itself will typically not be as probative of secondary meaning in a product configuration case as in a trademark case, since the product's market success may well be attributable to the desirability of the product configuration rather than the source-designating capacity of the supposedly distinguishing feature or combination of features. And unlike with a trademark, where repeated purchases of a product support an inference that consumers have associated the mark with the producer or source, one can much less confidently presume that a consumer's repeated purchase of a product has created an association between a particular product configuration and the source. Cf. International Jensen, 4 F.3d at 824 (product configuration) ("While evidence of a manufacturer's sales, advertising and promotional activities may be relevant in determining secondary meaning, the true test of secondary meaning is the effectiveness of this effort to create it."); Braun, Inc., 975 F.2d at 826-27 (product configuration) (same); First Brands Corp., 809 F.2d at 1383 (product packaging) (same); American Footwear Corp. v. General Footwear Co., 609 F.2d 655, 663 (2d Cir.1979) (trademark) (same).

The very fact that a consumer could identify the source based on the product's configuration implies that the configuration is at least somewhat unusual, and this "distinctiveness" of the product itself may be the source of the motivation to purchase if a consumer does not care about who the source is. In this respect product configuration again differs dramatically from trademark and from product packaging, since the success of a particular product--especially if similar competing products exist--does not readily lead to the inference of source identification and consumer interest in the source; it may well be that the product, inclusive of the product configuration, is itself inherently desirable, in a way that product packagings and trademarks are not. Similarly, unsolicited media coverage may reflect interest more in an unusual product than in the source of the product.

Analogously, attempts to copy a product configuration will quite often not be probative: the copier may very well be exploiting a particularly desirable feature, rather than seeking to confuse consumers as to the source of the product. See Aromatique, 28 F.3d at 871 (per curiam) (plurality opinion by Morris Sheppard Arnold, J.); Braun, Inc., 975 F.2d at 827; supra at 1446-47; cf. Brown, Design Protection: An Overview, 34 UCLA L. REV. at 1384-85 (observing that the act of copying a product configuration does not ipso facto involve false representation). The inference of unfair competition will be even weaker where the copier takes conspicuous steps--whether in packaging, trademark, marketing techniques, or otherwise--to distinguish its product from its competitor's.

In sum, secondary meaning in a product configuration case will generally not be easy to establish. We note, however, that in certain circumstances, e.g., with respect to drugs or pills with unusual colors and/or shapes, a consumer may be more likely to rely on the product's configuration as a source designator, and the consumer may thereby have become sensitized, so that secondary meaning will be easier to establish. This may be the case if the good is one with some features of importance to a consumer's choice that a consumer ordinarily cannot recognize in the market (such as the safety and efficacy of a drug) and one in which, because of the nature of the product's use or consumption, identifying source designations might not readily be prominently displayed. Cf. Sinko, 105 F.2d at 453 (stating that with products like drugs, where the efficiency of the product depends greatly upon the maker's capacity, consumers would care more about the pills' manufacturer than the appearance of the drug itself).

With respect to Duraco's Grecian Classics, we fully concur in the district court's finding of no secondary meaning. Duraco in its survey has not shown any consumer association between the Grecian Classics planters and a particular source; instead, its plastic planters are purchased because consumers (whether retail or wholesale) find them innately desirable, probably because of their pleasing "attic shape."14 Moreover, the evidence indicates without contradiction that Joy emulated Duraco's design because Joy believed it to be a superior one, not to trade on Duraco's non-existent good will in the Grecian Classics' configuration.

In addition, Duraco has not emphasized its alleged trade dress in its advertising, relying instead primarily on small depictions of the entire product. Finally, Duraco exclusively sold planters with the "Grecian Classics look" for about five years, not so long a time as to raise a strong inference of consumer association with a single source. Therefore, because the district court was correct and a fortiori not clearly erroneous, we will affirm its finding of no secondary meaning.

Applying the standard that this opinion has articulated for determining whether product configurations are inherently distinctive, we have concluded, viewing the evidence in the light most favorable to Duraco, that Duraco's Grecian Classics plastic planters are not inherently distinctive. We have further concluded that the district court was not clearly erroneous in finding that Duraco has failed to show secondary meaning. Thus, having demonstrated neither inherent nor acquired distinctiveness, Duraco has not established a likelihood of success on the merits of its Lanham Act action for trade dress infringement. For the foregoing reasons, we will affirm the district court's order denying Duraco's motion for a preliminary injunction. The district court's further proceedings resolving the case on final hearing will, of course, be conducted consistently with this opinion.