OAKES, Senior Circuit Judge.

This appeal raises the difficult question of under what circumstances the law of trade dress appropriately protects industrial design. Those who would seek monopolistic protection for industrial design, but are unable to obtain a design patent, have had little success under the law of copyright even with the “conceptual separateness” gloss provided by the 1976 Act. 1 Given impetus by the Supreme Court’s opinion in Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992), they have turned to the law of trade dress as it has evolved under the Lanham Act. When evaluating these claims, courts must not lose sight of the underlying purpose of the Lanham Act, which is protecting consumers and manufacturers from deceptive representations of affiliation and origin.

This is the second appeal of a preliminary injunction entered by the United States District Court for the Southern District of New York (John E. Sprizzo, Judge) in an action for trade dress infringement brought under § 43(a) of the Lanham Act, 15 U.S.C. § 1125(a), and state law. Plaintiff-Appellee Landscape Forms, Inc. (“Landscape”), a manufacturer of site furniture used in airports, parks, shopping malls and other similar locations, is incorporated in Michigan. Defendant-Appellant Columbia Cascade Co. (“Columbia”) competes with Landscape in the site furniture market and is incorporated in Oregon. On October 16, 1996, the district court entered an order reinstating an injunction previously signed on March 23, 1995, and filed on April 7, 1995, which prevented Columbia from “soliciting or accepting orders, delivering or selling” specified commercial, outdoor furniture. Columbia filed a Notice of Motion for a Stay of the Injunction Pending Appeal and a Notice of Appeal on November 8, 1996. A Motion for an Expedited Appeal was granted on December 3, 1996.

Landscape originally filed this lawsuit in November 1994 and then moved for a preliminary injunction on December 30, 1994. After several days of hearings, the district court enjoined Columbia from selling, in the United States, an entire line of its outdoor furniture which is similar to Landscape’s attractive “Petoskey” line.

The Petoskey line includes two different outdoor trash cans, two benches without back support, and six benches with backs. Of the benches in this last group, three stand on two legs, while three stand on two front legs and a pair of adjacent hind legs. All of the products, with the exception of one of the two trash cans, use curved steel pipe for support. The seating surfaces of the benches are wooden slats, steel rods, or curved, perforated metal sheets. All ten items are closely resembled by comparable Columbia products.

Our November 13, 1995, decision in the first appeal vacated the district court’s initial order enjoining Columbia and remanded this matter for consideration of Columbia’s affirmative defense that the design of Landscape’s furniture was functional. Landscape Forms, Inc. v. Columbia Cascade Co., 70 F.3d 251 (2d Cir.1995). That decision did not reach the issue whether there was a likelihood that consumers would be confused as to the identity of the manufacturer of the two product lines. Columbia now appeals the district court’s rejection of the functionality defense as well as the issues raised in its first *376 appeal. Citing Knitwaves, Inc. v. Lollytogs Ltd., 71 F.3d 996, 1008-09 (2d Cir.1995), Columbia also argues that Landscape’s designs do not qualify for trade dress protection because they are not “likely to serve primarily as source designators.” Our jurisdiction is based on 28 U.S.C. § 1292(a)(1). We agree with Columbia’s last contention and, therefore, once again vacate the injunction.

Background

A. District court findings prior to the first appeal.

On February 14, 1995, the district court issued an opinion from the bench. Without discussing whether consumers recognized Landscape as the manufacturer of the Petoskey furniture, the court found that Landscape’s designs were unique and distinctive. It then went on to apply the eight, nonexclusive factors from Polaroid Corp. v. Polarad Elecs. Corp., 287 F.2d 492, 495 (2d Cir.1961), to conclude that consumers were likely to be confused as to the identity of maker of the accused site furniture. 2 The district court found that the only factor “which may cut in favor of the defendant” was the sophistication of the relevant consumers who are contractors, landscape architects and other design professionals. But, in its discussion of other factors, the court mentioned several considerations favoring Columbia, such as the facts that each company’s products are labeled with the name of the manufacturer; that the benches were sold through catalogs that identify the manufacturer; and that there was little evidence of actual confusion. Among the factors supporting the injunction, the district court seemed most swayed by the strength of Landscape’s designs, which the court adjudged “unique, attractive and beautiful”; the similarity of the accused products’ overall look; and Columbia’s apparent bad faith in deliberately copying Landscape’s designs.

B. District court findings following our remand.

Following our remand for consideration of Columbia’s defense that the design of the site furniture was functional, and, thus, ineligible for trade dress protection, the district court heard from additional witnesses and made the following findings: First, “[wjhile it is unquestionably true that the Petoskey bench is an excellent and innovative design, it is clear from the evidence that there are a large number of available designs that can, with even a modest amount of creative effort, be utilized to compete effectively in terms of price, quality, and aesthetic appeal.” Landscape Forms, Inc. v. Columbia Cascade Co., 940 F.Supp. 663, 665 (S.D.N.Y.1996). Second, persuasive expert testimony established that several competitive designs already exist. Id. Third, Columbia failed to show that its overall business was adversely affected by the trade dress protection afforded to Landscape’s products; and, in fact, the accused line of site furniture was a “small fraction” of Columbia’s business. Id. at 666. The district court never found that Landscape’s designs were recognized as such by consumers in the marketplace, or, in other words, that they had acquired secondary meaning. Indeed, when asked by Landscape’s counsel to make such a finding, the court expressed doubt that there was “enough of a record” to do so.

Discussion

To obtain a preliminary injunction, a party must show (a) irreparable harm should the injunction not be granted, and (b) either (i) a likelihood of success on the merits, or (ii) sufficiently serious questions going to the merits and a balance of hardships decidedly in favor of the movant. Jeffrey Milstein, Inc. v. Greger, Lawlor, Roth, Inc., 58 F.3d 27, 31 (2d Cir.1995) (citation omitted). We review the district court’s grant of a preliminary injunction for an abuse of discretion. Judge Newman’s salutary discussion of this standard of review in Stormy Clime Ltd. v. ProGroup, Inc., 809 F.2d 971 (2d Cir.1987), explained:

*377 Judicial discretion normally implies the authority in a decision-maker to decide a disputed matter either way, or, in some circumstances, to varying degrees, as where a continuance may be granted for any one of several appropriate intervals. Discretion is said to be “abused” (“exceeded” would be both a more felicitous and correct term) when the decision reached is not within the range of decision-making authority a reviewing court determines is acceptable for a given set of facts.

Id at 974 (citing Friendly, Indiscretion About Discretion, 31 Emory L.Rev. 747, 773-78 (1982)).

Landscape’s assertion that § 43(a) of the Lanham Act prevents Columbia from copying the appearance of the Petoskey furniture is a claim of trade dress infringement. Such an action requires a plaintiff-manufacturer to show (a) that its trade dress is entitled to protection under the Act, and (b) that the defendant’s dress infringes on the plaintiffs dress by creating a likelihood of confusion. To be entitled to protection under the Act, plaintiffs trade dress must either be inherently distinctive or be shown to have acquired distinctiveness through “secondary meaning.” Two Pesos, 505 U.S. at 774, 112 S.Ct. at 2760; EFS Mktg., Inc. v. Russ Berrie & Co., 76 F.3d 487, 490 (2d Cir.1996).

If the plaintiff succeeds in this first part of the inquiry, the manufacturer of the accused product(s) can still avoid liability by showing that “the similar arrangement of features is functional.” Stormy Clime, 809 F.2d at 974. This analysis requires courts to balance the policy of protecting consumers from confusion against that in favor of free competition. Traditionally applied to prevent the recognition of trademark protection for useful features of a product, the functionality doctrine extends to the aesthetic aspects of a product’s design. In Wallace Int’l Silversmiths, Inc. v. Godinger Silver Art Co., 916 F.2d 76, 81 (2d Cir.1990), we held that such a defense requires a showing that trade dress protection would deprive competitors of alternative designs, and, thus, foreclose competition from the relevant market. See Restatement (Third) of Unfair Competition [hereinafter Restatement] § 17 cmt. e (1995) (“A design is functional because of its aesthetic value only if it confers a significant benefit that cannot practically be duplicated by the use of alternative designs.”). We start by examining whether Landscape has met its burden of showing that its line of products is protectable and that Columbia’s similar products are likely to confuse consumers. We do not reach Columbia’s defense that Landscape’s products are functional, which was the issue that prompted our remand of November 13, 1995, because we conclude that Landscape has yet to show that its line of furniture is protectable.

A. Protectable — inherent distinction.

Prior to the decision of the Supreme Court in Two Pesos, this court required a plaintiff to prove that its trade dress had acquired secondary meaning to obtain protection under the Lanham Act. See Murphy v. Provident Mut. Life Ins. Co., 923 F.2d 923, 927 (2d Cir.1990). Two Pesos, accepting without discussion that the decor of the plaintiffs chain of Mexican restaurants was distinctive, 505 U.S. at 770, 112 S.Ct. at 2758, held that inherently distinctive trade dress could be protected under federal law without proof of secondary meaning, but did not address the legal standard that should be employed to measure the inherent distinctiveness of trade dress. The Court, thus, left us and the other lower courts with the difficult task of constructing such a standard.

Judge Friendly’s familiar test for the inherent distinctiveness of trademarks in Abercrombie & Fitch Co. v. Hunting World Inc., 537 F.2d 4, 11 (2d Cir.1976) appeared a natural recourse. The Abercrombie test classifies verbal marks into four categories which run in a continuum: (1) generic, (2) descriptive, (3) suggestive, and (4) arbitrary or fanciful. Generic words or phrases may never serve as trademarks; descriptive marks, while not inherently distinctive, may be trademarks upon a showing of secondary meaning; and suggestive, arbitrary and fanciful terms are inherently distinctive trademarks even without a showing of secondary meaning.

Yet, as we observed in Knitwaves, it does not “make sense” to apply the Aber *378 crombie test directly to product design or configuration cases. Knitwaves, 71 F.3d at 1007. We agreed with the Third Circuit that, in contrast to verbal marks, one cannot meaningfully ask whether a product’s design features are generic or otherwise descriptive of the product itself, but we declined to follow that circuit’s application of a wholly-new, multi-pronged test to product design and configuration cases under § 43(a). Id. at 1009 n. 6 (discussing Duraco Prods., Inc. v. Joy Plastic Enters., Ltd., 40 F.3d 1431 (3d Cir.1994)). Instead, we simply asked whether the design was likely to be understood as an indicator of the product’s source. Knit-waves, 71 F.3d at 1008. The autumnal pattern on the front of plaintiff Knitwaves’s sweaters was not indicative of source, and, thus, did not warrant protection as trade dress under federal law. We drew this requirement from “what [the Court] called, in Qualitex, ‘the more important part of the statutory definition of a trademark’ — the requirement ‘that a person “us[e]” or “inten[d] to use” the mark “to identify and distinguish his or her goods ... from those manufactured or sold by others and to indicate the source of the goods.” ’ ” Id. (quoting Qualitex Co. v. Jacobson Prods. Co., 514 U.S. 159, -, 115 S.Ct. 1300, 1303, 131 L.Ed.2d 248 (1995) (quoting 15 U.S.C. § 1127)). 3

Very recently, we distinguished Knitwaves and held that the Abercrombie test will still be applied to measure the distinctiveness of a product’s packaging in trade dress cases under the Lanham Act. Fun-Damental Too, Ltd. v. Gemmy Indus. Corp., 111 F.3d 993, 1000-01 (2d Cir.1997), concerned the intentional copying of several aspects of a box used to contain and display the “Toilet Bank,” a novelty coin bank, by the manufacturer of a similar novelty product, the “Currency Can.” The decision distinguished product packaging trade dress cases from product configuration cases, such as Knit-waves and the case at bar, for several reasons: First, we expressly held in Paddington *379 Corp. v. Attiki Importers & Distribs., Inc., 996 F.2d 577, 588 (2d Cir.1993), that Abercrombie applies to product packaging cases. Fun-Damental Too, 111 F.3d at 1000-01. Second, while it is difficult to define some aspect of a product’s design or configuration in terms of the Abercrombie categories, “a product’s packaging style creates an image of the product more readily separated from the product itself.” And, “although there may be a finite set of ways to configure a product, the variety of packaging available for a given product is limited only by the bounds of imagination.” Id. at 1000-01. Third, “applying Abercrombie to product packaging serves the aims of the Lanham Act because consumers are more likely to rely on the packaging of a product than on the product’s design as an indication of source.” Id. at 1001 (citing Restatement § 16 cmt. b). For these reasons, Fun-Damental Too limited Knitwaves to its facts, and held that only when trade dress is claimed in the configuration or design of the product itself will the test for inherent distinction be whether the claimed dress is “likely to serve as a source designator.”

The case at bar clearly falls under Knit-waves as it concerns only the design of a series of products, not their packaging. Nonetheless, we digress momentarily to discuss Knitwaves and Fun-Damental Too in light of the purpose of the Lanham Act. We do so because we are concerned that the district courts may be unnecessarily confused by the two eases, which appear to divide the federal law of trade dress into two distinct realms, each with its own test for inherent distinction. We are not so confident that the Abercrombie analysis is more naturally fit for direct application to product packaging cases than configuration cases, but we nevertheless agree with the differences noted in FunDamental Too between packaging and configuration cases. Indeed, judging from that decision, this circuit appears to be moving toward a rule that packaging is usually indicative of a product’s source, while the design or configuration of the product is usually not so. See 111 F.3d at 1000 (“[T]he varieties of labels and packaging available to wholesalers and manufacturers are virtually unlimited. As a consequence, a product’s trade dress typically will be arbitrary or fanciful and meet the inherently distinctive requirement for § 43(a) protection.”) (citations omitted). Neither principle is hard and fast, and analysis will always require a look at the product and the market in which it competes. For example, in Mana Prods., Inc. v. Columbia Cosmetics Mfg., Inc., 65 F.3d 1063, 1070 (2d Cir.1995), we held that black, compact containers for cosmetics were so common in that industry as to be generic and, hence, not protectable. Application of these principles will also be complicated by the fact that it may at times be difficult to say exactly where a product stops and its packaging begins; this was arguably the case in Mana Prods. where the makeup compact, which is not discarded after purchase, could be considered to be either the product or part of its packaging.

Whether or not the Abercrombie analysis is employed, the Lanham Act must be construed in the light of a strong federal policy in favor of vigorously competitive markets, which is exemplified by the Sherman Act and other anti-trust laws. This fundamental principle is reflected in decisions holding that federal patent law preempts state unfair competition laws that limit competitors’ use of inventions in the public domain. See, e.g., Compco Corp. v. Day-Brite Lighting, Inc., 376 U.S. 234, 84 S.Ct. 779, 11 L.Ed.2d 669 (1964); Sears, Roebuck & Co. v. Stiffel Co., 376 U.S. 225, 84 S.Ct. 784, 11 L.Ed.2d 661 (1964). While the Lanham Act is not “preempted” by other federal statutes in the same way as was the state law in Sears and Compco, section 43(a) complements the federal policy favoring competition. “Congress recognized that a trademark aids competition in the marketplace because it helps a consumer distinguish among competing products. Trademarks also encourage producers to maintain a high quality product by assuring that any goodwill associated with their products is not misappropriated by competitors.” Fun-Damental Too, 111 F.3d at 999 (citing S.Rep. No. 79-1333 (1946), reprinted in. 1946 U.S.C.C.A.N. 1274,1275). The functionality defense, moreover, by denying an otherwise valid claim of trade dress infringement if its recognition would create an un *380 warranted monopoly, protects competition even at the cost of potential consumer confusion. See Qualitex, 514 U.S. at -, 115 S.Ct. at 1306; Inwood Lab., Inc. v. Ives Lab., Inc., 456 U.S. 844, 850 n. 10, 102 S.Ct. 2182, 2187 n. 10, 72 L.Ed.2d 606 (1982); Kellogg Co. v. National Biscuit Co., 305 U.S. 111, 122, 59 S.Ct. 109, 115, 83 L.Ed. 73 (1938).

A similar concern for the protection of competition informs the initial stage of the analysis of a claim of trademark or trade dress infringement, i.e., determination of whether a mark, packaging or product design qualifies for protection. The Abercrombie analysis illustrates this underlying concern of the Lanham Act: Generic terms cannot be protected because, inter alia, competitors need them to describe their products to consumers. Id. at 9 (“[T]he user of a generic term ... cannot deprive competing manufacturers of the product of the right to call an article by its name.”). And, on the other end of the spectrum, arbitrary or fanciful words and phrases may be appropriated by a single producer without depriving others of a means of describing their products to the market. Cf. Paddington, 996 F.2d at 584 (liquor bottle label was per se distinctive because design elements were selected from “an almost limitless supply of patterns, colors and designs”). The Abercrombie test, thus, permits courts to separate those cases in which similar marks are most likely to mislead consumers from those in which trademark protection would create a “linguistic monopoly” which would stifle competitors’ efforts to market similar goods to consumers.

The policy of protecting competition is at least as strongly implicated when, as in the present case, product designs or configurations are claimed as trade dress. While trademarking a generic term would create a monopoly in a necessary word or phrase, granting trade dress protection to an ordinary product design would create a monopoly in the goods themselves. For this reason, courts have exercised particular “caution” when extending protection to product designs. Jeffrey Milstein, 58 F.3d at 32. Accordingly, we think that our decision in Knitwaves was consistent with the policy of Abercrombie, which is to accord protection to symbols consumers are likely to rely upon in distinguishing goods, while denying protection that would hamper efforts to market competitive goods. Viewed in this light, the difference between Knitwaves and Fun-Damental Too is not great, as both “tests” seek answers to the same questions in aid of advancing the same underlying policy.

Judge Newman articulated similar concerns in Jeffrey Milstein. Jeffrey Mil-stein declined to extend trade dress protection to a series of greeting cards with “ ‘straight-on, strong photographic, glossy images of animals, persons or objects on die-cut cards____’ ” 58 F.3d at 33 (citation omitted). In rejecting this claim for protection of “an idea or concept” that is too broad or too general to warrant protection, Judge Newman made two points pertinent to the present appeal: first, “overextension of trade dress protection can undermine restrictions in copyright and patent law that are designed to avoid monopolization of products and ideas”; and “[sjecond, just as copyright law does not protect ideas but only their concrete expression, neither does trade dress law protect an idea, a concept, or a generalized type of appearance.” Id. at 32.

Jeffrey Milstein also demonstrates that a trade dress plaintiff seeking to protect a series or line of products faces the particularly difficult challenge of showing that the appearance of its several products is sufficiently distinct and unique to merit protection as a recognizable trade dress. Compare Walt Disney Co. v. Goodtimes Home Video Corp., 830 F.Supp. 762, 766 (S.D.N.Y.1993) (discussing the special burden when claim is for the “overall look of a number of different packages”) with Harlequin Enters. Ltd. v. Gulf & Western Corp., 644 F.2d 946, 948 (2d Cir.1981) (“every book in the Harlequin Presents series is uniform in its dimensions, cover design and colophon”). Furthermore, a claim of trade dress covering an array of items is likely to be broader than one for an individual product’s design. Accordingly, when protection is sought for an entire line of products, our concern for protecting competition is acute. Cf. Wallace, 916 F.2d at 81-82 (claim of trade dress in baroque-style decorations on silverware conflicted with the *381 aesthetic functionality doctrine, because it was too broad).

Saying this, we still recognize that there is no question that trade dress may protect the “overall look” of a product. See Bristol-Myers Squibb Co. v. McNeil-P.P.C., Inc., 973 F.2d 1033, 1042 (2d Cir.1992). Thus, “[a]lthough each element of a trade dress individually might not be inherently distinctive, ... the combination of elements” may be indicative of source. Jeffrey Mil-stein, 58 F.3d at 32. Nonetheless, focus on the overall look of a product does not permit a plaintiff to dispense with an articulation of the specific elements which comprise its distinct dress. Without such a precise expression of the character and scope of the claimed trade dress, litigation will be difficult, as courts will be unable to evaluate how unique and unexpected the design elements are in the relevant market. Courts will also be unable to shape narrowly-tailored relief if they do not know what distinctive combination of ingredients deserves protection. Moreover, a plaintiffs inability to explain to a court exactly which aspects of its product design(s) merit protection may indicate that its claim is pitched at an improper level of generality, i.e., the claimant seeks protection for an unprotectable style, theme or idea.

Under this test, Landscape has not articulated and supported its claimed trade dress with sufficient particularity. Averring that it “seeks protection for a specific expression of site furniture ...,” Landscape now explains to this court: “The Petoskey trade dress incorporates large three-inch tubing, with a powdered cosmetic finish, bent in gentle turns that roll around the perimeter of the furniture which in combination with the various seating surfaces gives the viewer a floating or suspended feeling.” The complaint, which takes aim at Columbia’s copying of Landscape’s products, contains no similar description of the alleged dress. Paragraph 11 states “[t]he product design and configuration of [Landscape’s] Petoskey" Group Collection employs a number of distinctive elements which, when taken together, constitute a trade dress recognizable by architects, landscape architects and designers, as well as the public at large.” The “distinctive elements” are not enumerated.

A statement by Landscape’s attorney and testimony from some witnesses started to articulate the unique qualities of the company’s line of products. In her opening statement, Landscape’s attorney explained:

The furniture has been described as actually floating or suspended. It incorporates large, round 3-inch-diameter pipes which are bended and curved—

giving it a wavelike effect. It uses rods for its seating, round rods, and in some cases uses perforated metal in the seating to give a see-through effect.

William Main, Landscape’s President, added:

The Petoskey Dine] gets a lot of its look from the bold tube, powder coated tube that comes up from the ground and kind of flows around, bends and flows around the back of the bench of the litter receptacle, and there is a series of different materials that can be placed in that support to sit on, that have kind of a floating feel. The look is kind of cantilevered, and it is very unexpected.

And, at the remand hearing, Amo Yurk, an industrial designer at Landscape, testified:

There’s a lot of curves and soft corners that go on. There’s a lot of structural heft when you look at it. It seems to float in this rather heavy-looking framework, and I think in that regard all the pieces seem to tie together, the common thread being, in my mind, that very heavy structure that sort of lets the elements float inside of there.

Others echoed Yurk’s sentiment that the Petoskey furniture is “substantial” yet appears “light.” Indeed, the photographs of the furniture capture this pleasantly paradoxical feel of some of Landscape’s benches.

Yet, these and the many other similarly laudatory descriptions in the record fail to indicate what unique combination of features makes the trade dress of the ten items in the Petoskey line inherently distinctive under the Knitwaves test, i.e., likely to be perceived by consumers as bearing the stamp of their *382 maker. A claim for site furniture which is at once massive, yet appears to float, is too abstract to qualify as trade dress for the reasons discussed in Jeffrey Milstein. If the law protected style at such a level of abstraction, Braque might have prevented Picasso from selling cubist paintings in the United States.

Tellingly, the more detailed parts of the above descriptions apply to some, but not all, of the ten items in the Petoskey line. Three of the six benches with backs are supported by two legs made from a single pipe which “flows around” the “perimeter” and are thus “cantilevered.” The trash cans, backless benches, and three-legged benches, however, do not readily fit that description. For example, the two, rather pedestrian “U” shaped tubes, which attach to either end of the backless benches for support, cannot be said to “flow” in the same way as the tubes on any of the benches with backs. The use of bent, three-inch pipe or tube is one element of the designs which arguably unites them, yet this feature alone cannot support Landscape’s broad claim for trade dress protection. First, one of the two trash cans includes no such pipe. Second, as just mentioned, not all the benches use the bent pipe in the same way, or to the same effect. Third, and most importantly, even Landscape’s president testified that bent tubes are commonly used in outdoor furniture. Our browse through the portions of industry catalogs included in the record confirms this admission. Accordingly, we do not think that this design element alone can qualify the entire Petoskey line as inherently distinctive trade dress.

In sum, Landscape’s proof at the proceedings below did not focus on the specific features of its line of furniture that the company claims as trade dress. Without more exacting analysis, any new product — particularly a successful one — could conceivably qualify as inherently distinctive. Landscape’s failure is especially troubling because it sought to protect an entire array of products. We hold that the district court exceeded its decision-making authority when it concluded that Landscape demonstrated that its Petoskey line of products could be protected as trade dress without proof of secondary meaning. None of which is to say that we disagree with the proposition expressed some years ago in a concurring opinion of Justices White and Marshall that “[t]he use of a product or package design that is so similar to that of another producer that it is likely to confuse purchasers as to the product’s source may constitute ‘false designation of origin’----” Inwood Lab., 456 U.S. at 863, 102 S.Ct. at 2193 (White, J., concurring). We think, however, that Landscape has yet to articulate what makes its products inherently distinctive such that, without proof of secondary meaning, there is reason to believe that confusion is likely.

B. Likelihood of confusion.

In light of our conclusion above, we do not need to address the district court’s likelihood of confusion analysis to resolve the Lanham Act claim. Likelihood of confusion is relevant, however, to state law, and further elucidation of the issue may be helpful to the district court. We, therefore, briefly discuss the district court’s analysis.

The district court afforded too much weight to the first of the Polaroid factors. The district court concluded that Landscape had not demonstrated secondary meaning, and, for reasons already discussed, its conclusion that the several designs were inherently, strongly indicative of a particular source was error. Second, Columbia and Landscape are clearly identified as the manufacturers of their respective lines of furniture in industry catalogs, and their products bear labels identifying their maker. These marketplace realities make confusion improbable, even though the two product lines compete in the same market. Third, although the district court recognized the sophistication of the design professionals who are the consumers in the site furniture market, it did not give this factor the importance it deserves. The district court said that the risk of after-market confusion limited the importance of the consumer sophistication factor. We disagree. The likelihood of confusion test concerns not only potential purchasers but also the general public. See United States v. Hon, 904 F.2d 803, 804-08 (2d Cir.1990) (discussing Rolex watches and other luxury goods). But, such third parties are only relevant if their views are somehow *383 related to the goodwill of the aggrieved manufacturer. Restatement § 20 cmt. b. Here, where there is no showing that the general public is aware of Landscape’s “dress,” the district court erred in giving this factor great weight. Last, simulating the design of a competitor’s successful products is not bad faith, unless there is reason to draw an inference of an intention to deceive. Restatement § 22 cmt. c (“It is the intent to cause confusion or to deceive that may justify an inference that confusion is likely.”); cf. My-T Fine Corp. v. Samuels, 69 F.2d 76, 77 (2d Cir.1934) (Learned Hand, J.). The record demonstrates no such intent.

The district court, thus, erred in its conclusion that Landscape demonstrated a likelihood of confusion.

C. New York law.

Landscape initiated this proceeding under both the Lanham Act and New York law. 4 The district court did not rely on state law when it issued the injunction, nor was state law briefed to this court. We note that reliance on state law to protect these ten product designs would raise a serious question of preemption by federal design patent law under the Sears-Compco doctrine, even as it has been narrowed in Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141, 154, 109 S.Ct. 971, 979, 103 L.Ed.2d 118 (1989). Mindful of the still early stage of this proceeding, we are wary of deciding a difficult question of federalism which has not been properly raised or briefed. Fortunately, our conclusion that Landscape has not shown a likelihood of consumer confusion disposes of the problem at this stage, as New York law requires such a showing for equitable relief. Jeffrey Milstein, 58 F.3d at 35 (citing W.W.W. Pharm. Co. v. Gillette Co., 984 F.2d 567, 576 (2d Cir.1993)).

Conclusion

Injunction vacated.

. See, e.g., Erica Lehrer, The Design of Design Law Today, 35 Copyright L.Symp. (ASCAP) 145 (1988); Robert C. Denicola, Applied Art and Lndustrial Design: A Suggested Approach to Copyright in Useful Articles, 67 Minn.L.Rev. 707 (1983).

. The eight factors are: "the strength of [the] mark, the degree of similarity between the two marks, the proximity of the products, the likelihood that the prior owner will bridge the gap, actual confusion, and the reciprocal of defendant’s good faith in adopting its own mark, the quality of defendant's product, and the sophistication of the buyers.” Polaroid, 287 F.2d at 495.

. Knitwaves has been followed in this court by EFS Mktg. and at the district court level by Herbko Int’l, Inc. v. Gemmy Indus. Corp., 916 F.Supp. 322, 328 (S.D.N.Y.1996); Grupke v. Linda Lori Sportswear, Inc., 921 F.Supp. 987, 996 (E.D.N.Y.1996); and Banff Ltd. v. Express, Inc., 921 F.Supp. 1065, 1070-71 (S.D.N.Y.1995).

In Krueger Int’l, Inc. v. Nightingale Inc., 915 F.Supp. 595 (S.D.N.Y.1996), however, Judge Sotomayor criticized Knitwaves for requiring courts to decide whether the primary purpose of trade dress is either aesthetic or source-identifying. She rightly observed that trade dress is usually meant to please. 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.

Knitwaves declined to extend trade dress protection to ornamentation on the front of sweaters where there was no reason to believe that the designs were likely to be understood as an indication of the manufacturer of the sweaters. The case turned entirely upon the lack of inherent distinctiveness of the patterns, because there was no showing that they had achieved any secondary meaning in the market for children's clothing. Because evaluation of inherent distinctiveness involves a more confined analysis than the inquiry into whether dress has gained secondary meaning, caution is needed to prevent imprudent recognition of trade dress in product features that could not plausibly be taken to indicate the manufacturer. Judge Sotomayor employed the series of questions posed in Seabrook Foods, Inc. v. Bar-Well

Foods Ltd., 568 F.2d 1342, 1344 (C.C.P.A.1977) to assess whether the design of the Matrix model of high-densily, stacking chairs is inherently distinctive. In her discussion, she indicates that the Seabrook "test” is inconsistent with Knitwaves. We disagree. The questions posed in Seabrook were "whether [the design] was a ‘common’ basic shape or design, whether it was unique or unusual in a particular field, [or] whether it was a mere refinement of a commonlyadopted and well-known form of ornamentation for a particular class of goods” which consumers view as such. Id. These questions may, in different contexts, be useful tools to assess whether a design is “likely to be perceived as a source indicator,” See 1 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition §§ 8-13 (4th ed. 1996) ("In reality, all three [of the Seabrook] 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.” (footnote omitted)). Indeed, Knitwaves fits neatly under the last Seabrook question, as the autumnal pattern is a "commonly-adopted ... form of ornamentation.”

As indicated in Knitwaves, a manufacturer’s subjective intentions may be probative of whether its dress is likely to indicate product source. See Knitwaves, 71 F.3d at 1008-09 (referring to manufacturer's intent). But, as suggested by Seabrook, objective consideration of the product and its similarity to others on the market will always be relevant and often decisive.

. In addition to the state unfair competition claim, Landscape’s complaint alleged dilution under New York law. We do not address this theory which was neither litigated below nor briefed to this court.