This case requires us to interpret and apply the dauntingly elusive concept of trademark "dilution" as now embodied in the Federal Trademark Dilution Act of 1995 ("the Act"). See Federal Trademark Dilution Act of 1995, Pub.L. No. 104-98, 109 Stat. 985 (codified at 15 U.S.C. §§ 1125, 1127). The concept was invoked in this case by Ringling Bros.-Barnum & Bailey Combined Shows, Inc. ("Ringling") in a claim under the Act that Ringling's "famous" circus trademarks logan, THE GREATEST SHOW ON EARTH ("GREATEST SHOW mark"), had been diluted by the State of Utah's commercial use of its trademark slogan, THE GREATEST SNOW ON EARTH ("GREATEST SNOW mark"), as an advertisement of the state's winter sports attractions. The district court found that Ringling had not proved dilution under the Act and gave judgment for Utah. We affirm the judgment.

* The relevant background facts as found by the district court are undisputed. From 1872 to the present, Ringling and its predecessors have offered their circus to the public as the "Greatest Show on Earth." In 1961, Ringling received federal trademark registration for its GREATEST SHOW mark for entertainment services in the nature of a circus.

Since its inception, Ringling has used its mark to advertise circus performances. The circus travels throughout the United States and presents approximately 1,000 shows annually to some 12 million people in 95 cities. More than 70 million people each year are exposed to the GREATEST SHOW mark in connection with the circus. Revenues derived from goods and services bearing or using the mark are substantial and exceeded $103 million for the fiscal year ending January, 1997.

Ringling advertises its circus using the GREATEST SHOW mark in print advertising, radio, television, videos, outdoor billboards, direct-mail pieces, press announcements, posters, program books, souvenirs, and joint promotions with other companies. In the fiscal year ending January 1997, expenditures on advertising using the mark totaled approximately $19 million. Through joint promotions with retailers, Ringling obtains significant additional exposure for its mark. Also, because of its renown, the GREATEST SHOW mark receives substantial free publicity.

Defendant Utah Division of Travel Development ("Utah") is an agency of the State of Utah. As early as 1962, Utah began using its GREATEST SNOW mark in connection with Utah tourism services. Utah has used its mark in magazine advertisements every year from 1962 to the present except 1963, 1977, and 1989. Utah has authorized the Utah Ski Association to use the GREATEST SNOW mark in connection with the Association's promotion of Utah tourism. Utah's primary use of its mark in Utah is its display on motor vehicle license plates. For each of the past fifteen years, Utah's budget for winter advertising, including advertising of the GREATEST SNOW mark, has ranged from $300,000 to $450,000.

In 1965, the Utah Attorney General opined that Utah's mark did not impair or violate Ringling's GREATEST SHOW mark. Utah registered its mark with the State of Utah in 1975 and renewed its registration in 1985 and 1995. In 1988, Utah applied to the United States Patent and Trademark Office to register its mark. Although Ringling opposed Utah's application, Utah was granted federal registration for its mark on January 21, 1997.

On June 6, 1996, Ringling commenced this action, seeking injunctive and monetary relief, on allegations that Utah's use of the GREATEST SNOW mark "diluted" Ringling's GREATEST SHOW mark in violation of the Act. Before trial, the district court granted Utah's motion to strike Ringling's demand for a jury trial, and after a bench trial, found for Utah.

This appeal by Ringling followed. Before us, Ringling challenges the district court's determination on the merits that Utah's mark did not dilute Ringling's mark in violation of the Act, and the court's denial of its demand for jury trial. We take these in turn.

The Federal Trademark Dilution Act, which became effective on January 16, 1996, amended Section 43 of the Lanham Act to provide a new cause of action for federal trademark "dilution." Under the Act, the owner of a "famous mark" is given protection "against another person's commercial use ... of a mark or trade name, if such use begins after the mark has become famous and causes dilution of the distinctive quality of the mark." 15 U.S.C. § 1125(c)(1). A successful claimant may be given injunctive and, if a willful violation is proved, restitutionary, compensatory, and specific relief in the form of a destruction of offending articles. See id. §§ 1125(c)(1)-(2), 1117(a),1118.

the lessening of the capacity of a famous mark to identify and distinguish goods or services, regardless of the presence or absence of--

(1) competition between the owner of the famous mark and other parties, or

(2) likelihood of confusion, mistake, or deception.

Id. § 1127.

And, the Act's legislative history further indicates that Congress understood that "dilution" might result either from "uses that blur the distinctiveness of [a famous] mark or [that] tarnish or disparage it." See H.R.Rep. No. 104-374, at 2 (1995), U.S. Code Cong. & Admin. News at 1029, 1029. The parties here both accept this as a proper reflection of congressional intent respecting the meaning of "dilution," and further agree that only dilution by blurring is at issue in this case.1

To prove its statutory dilution claim, Ringling's burden therefore was to prove (1) that its mark was a "famous" one; (2) that Utah adopted its mark after Ringling's had become famous; and (3) that Utah's mark diluted Ringling's by "blurring" it. See id. §§ 1125(c)(1), 1127.

At trial, Ringling put on essentially undisputed evidence demonstrating that its mark had achieved "famous" status before Utah began use of its mark. This left as the dispositive issue whether Utah's mark had "diluted" Ringling's by "blurring" it. On that issue, Ringling took the position that as a matter of statutory interpretation, "dilution" by "blurring" occurs whenever a junior mark is either identical to or sufficiently similar to the famous mark that persons viewing the two instinctively will make a "mental association" between the two. (Appellant's Br. at 9.) On this interpretation, viewers' knowledge of the goods or services represented by the two marks is irrelevant; all that counts is the identity or sufficient similarity of the marks as perceived by the viewer. Taking this as the legal meaning of "dilution by blurring," Ringling then contended that, though not identical, the similarity between Ringling's and Utah's marks was so strong and obvious that the required "mental association" of the two, hence the dilution by blurring of Ringling's senior mark, was evident as a matter of law, no other evidence being required to establish it. But, in a back-up position, Ringling presented evidence of a survey of hypothetical viewers designed to demonstrate empirically that the two marks did evoke in a properly representative sampling of viewers the mental association of the two that sufficed alone to prove a dilution violation.

In a comprehensive opinion, see Ringling Bros.-Barnum & Bailey Combined Shows, Inc. v. Utah Div. of Travel Dev., 955 F. Supp. 605 (E.D.Va.1997), the district court rejected Ringling's critical legal contentions respecting the legal meaning of "dilution" as used in the Act and found its attempted factual proof of dilution by means of a viewer survey insufficient to establish a violation.

Specifically, the court rejected Ringling's contention that proof alone of an "instinctive mental association" of the two marks by viewers sufficed to prove "dilution." While recognizing that to prove dilution by blurring one must necessarily prove as a threshold element a mental association by viewers of the marks themselves, the court held that this alone did not suffice. Rather, the court held, dilution by blurring occurs only where consumers "mistakenly associate or confuse the marks and the goods or services they seek to identify and distinguish," and this association causes actual harm to the senior mark's capacity to "identify and distinguish." Id. at 615-16. Applying this interpretation of "dilution" to Ringling's consumer survey evidence, the court found that the attempted proof by this means failed. See id. at 616-18. And, finally, analyzing the evidence as a whole under a multi-factor balancing test proposed for the purpose in Mead Data Central, Inc. v. Toyota Motor Sales, U.S.A., Inc., 875 F.2d 1026, 1035 (2d Cir.1989) (Sweet, J., concurring), the court concluded that "dilution" had not been established on a balancing of those factors. See id. at 618-22.

On this appeal, Ringling challenges both the district court's interpretation of the statutory meaning of "dilution"; the court's rejection of its survey evidence as insufficient to prove "dilution"; and the court's rejection of its dilution claim under a Mead Data analysis. We take these in turn.

Ringling's primary challenge is to the district court's interpretation of the statutory meaning of "dilution," hence of the elements of the "dilution" claim newly created by the Act. As in the district court, it contends for its contrary interpretation: that a famous mark is "diluted" whenever a junior mark is sufficiently similar that consumers viewing them "instinctively make a mental association" of the two. It therefore argues that the district court erred in interpreting the Act to require further proof that in making this "mental association" consumers "mistakenly associate or confuse the marks and the goods or services they seek to identify and distinguish." (Appellant's Br. at 9, 18-20.) And, it further argues that the court erred in interpreting the Act to require proof of "actual dilution." (Appellant's Reply Br. at 3-4.)

Reviewing de novo the statutory interpretation issue, see Shafer v. Preston Memorial Hosp. Corp., 107 F.3d 274, 277 (4th Cir.1997), we disagree with Ringling's proffered "mental-association-alone" interpretation. Though we do not agree in every particular with the district court's interpretation, we agree with its basic points that "dilution" under the federal Act consists of (1) a sufficient similarity of marks to evoke in consumers a mental association of the two that (2) causes (3) actual harm to the senior marks' economic value as a product-identifying and advertising agent.

That meaning surely does not leap fully and immediately from the statutory text. But, we believe it is the necessary meaning of the Act's critical provisions when read in light of the Act's legislative history. By that, we mean both the immediate but quite meager legislative record and, more critically, the broader background out of which the basic concept emerged and has evolved in state and federal trademark law. Though the process is laborious, we believe the historical inquiry has to begin far back with intellectual origins.

The concept of trademark "dilution" as distinct from "infringement" is commonly traced (though there were exploratory judicial antecedents) to Frank I. Schechter. See Restatement (Third) of Unfair Competition § 25 cmt. b. (1995) [hereinafter, Restatement ]; see also Mead Data Central, 875 F.2d at 1028; Norm Thompson Outfitters, Inc. v. General Motors Corp., 448 F.2d 1293, 1299 (9th Cir.1971). Pointing up the inadequacies of then-current trademark law to serve its consumer-protection function in the complex, multi-layered marketing systems that had now evolved, Schechter first proposed simply expanding the protections provided by the consumer-protection model to accommodate the new market realities. See Frank I. Schechter, The Historical Foundations of the Law Relating to Trade-Marks (1925).

Two years later however, Schechter had concluded and advanced the thesis that the consumer-protection model, even in its expanded state, could not adequately accommodate the realities of twentieth century marketing. See Frank I. Schechter, The Rational Basis of Trademark Protection, 40 Harv. L. Rev. 813 (1927) [hereinafter Schechter, Rational Basis of Trademark Protection ]. His proposal now was to abandon that model entirely, recognize that "the preservation of the uniqueness of a trademark ... constitute[s] the only rational basis for its protection," id. at 831, and provide that protection by prohibiting "dilution" of such a mark's uniqueness from which it derived its hard-earned advertising value and selling power. See id. at 832 (borrowing the term "diluted" from a German case). Under that proposal, trademark law would have been confined to preventing the "dilution" of truly "unique" marks identified as those employing "coined, arbitrary or fanciful words or phrases that have ... from the very beginning, been associated in the public mind with a particular product." Id. at 829. And, by "dilution" under this model was meant simply any junior use of an identical or sufficiently similar mark, without regard to whether the junior use had any other harmful effect than its necessary destruction of the senior mark's former absolute "uniqueness" as a product symbol. See id. at 825. This flowed from Schechter's thesis that the "real injury" caused by concurrent use of such marks was not consumer confusion but "the gradual whittling away or dispersion of the identity and hold upon the public mind of the mark or name by its use upon non-competing goods." Id. Protection of the public against deceptive and confusing uses of non-"unique" marks would have been left under such a regime to other laws than that of trademark.

This radical dilution proposal, whose practical effect if fully adopted would be to create as the whole of trademark-protection law property rights in gross in suitably "unique" marks, never has been legislatively adopted by any jurisdiction in anything approaching that extreme form. In fact, though from the outset its basic concept evoked occasional favorable judicial notices, see, e.g., Tiffany & Co. v. Tiffany Prods., Inc., 147 Misc. 679, 264 N.Y.S. 459, 462 (N.Y.Sup.Ct.), aff'd, 237 A.D. 801, 260 N.Y.S. 821 (1932), aff'd, 262 N.Y. 482, 188 N.E. 30 (1933), there was no legislative adoption of the concept in any form until 1947 when Massachusetts enacted the first state "antidilution" statute. See Act of May 2, 1947, ch. 307, § 7a, 1947 Mass. Acts 300 (codified as amended at Mass. Gen. Laws Ann. ch. 110B, § 12 (West 1996)). Over the next fifty years, other states followed suit and by 1996, when the President signed the federal Act into law, around half of the states had done so. See Restatement, supra § 25 statutory note. Though they of course varied in detail, the state statutes typically had four features of relevance for our interpretive purposes: (1) they defined the category of marks protected against dilution solely by reference to their "distinctive quality"; (2) they proscribed not just actual, consummated dilution, but the "likelihood of dilution"; (3) by containing no express reference to harm to the senior mark's economic value, they defined dilution in terms susceptible to the interpretation that it consisted solely of a loss of the mark's distinctiveness; and (4) they provided only injunctive relief. See, e.g., N.Y. Gen. Bus. Law § 360-1 (McKinney 1998); Ala.Code § 8-12-17 (1998); Cal. Bus. & Prof.Code § 14330 (West 1998).

It was against this background that Congress in 1995 brought the dilution concept into federal trademark law by the Federal Trademark Dilution Act's amendment of the Lanham Act. See 15 U.S.C. §§ 1125(c), 1127.2 Though the sparse congressional record of the amendment's adoption contains no allusion to this background, its principal features must necessarily have figured in Congress's understanding and purpose in adopting the concept almost seventy years after it was first proposed in theoretical form and almost fifty years after the states had first brought it into state trademark law. That history, consisting of the concept as first proposed, the states' legislative adoption of modified forms, and the courts' reaction to the eventual state antidilution legislation, provided the sole primary sources for congressional understanding of the concept and purpose for adopting it in any form.

The most critical aspect of that history for our purposes is the experience of courts in interpreting and applying the state antidilution statutes. The broad outlines of that experience are fairly summarized by the Restatement on Unfair Competition in a comment to its section on "Dilution and Tarnishment":

At first the courts applied the statutes reluctantly, if at all. In many cases dilution claims were denied because the plaintiff failed to prove a likelihood of confusion, notwithstanding the clear language of the statutes eliminating confusion as an element of the cause of action. Some courts, and numerous commentators, expressed fear that the uncertain limits of the antidilution cause of action would unduly expand the ability of trademark owners to monopolize language and inhibit free competition. A broad antidilution theory also has the potential to render superfluous the traditional likelihood of confusion standard of liability. It was further suggested that a state cause of action for dilution might interfere with the federal policy of uniform, national trademark protection implemented under the Lanham Act, although only isolated cases supported this preemption theory. Other commentators, however, continued to urge protection for the selling power of well-known trademarks. After the New York Court of Appeals in Allied Maintenance Corp. v. Allied Mechanical Trades, Inc., 42 N.Y.2d 538, 399 N.Y.S.2d 628, 369 N.E.2d 1162 (1977), expressed the need for protection against the "cancer-like growth of dissimilar products or services which feeds upon the business reputation of an established distinctive trade-mark or name," judicial acceptance of the antidilution statutes increased. Nonetheless, in apparent recognition that broad interpretation of the statutes would undermine the balance between private and public rights reflected in the traditional limits of trademark protection, the courts have continued to confine the cause of action for dilution to cases in which the protectable interest is clear and the threat of interference is substantial.

Restatement, supra § 25 cmt. b.

Within that general experience, several features are of particular relevance to our purpose. First is the sheer difficulty that courts have had in getting a firm handle on the basic concept of "dilution" as cryptically expressed in the typical state statute in an unelaborated reference to "dilution of the distinctive quality of a mark." E.g., N.Y. Gen. Bus. Law § 360-1. That the difficulty has been a significant one is sufficiently demonstrated by the wavering course of judicial reactions noted in the Restatement comment. That it has persisted is sufficiently captured by the Second Circuit's plaintive observation in 1983, more than thirty years after courts first began grappling with the interpretive problem, that "dilution remains a somewhat nebulous concept." Sally Gee, Inc. v. Myra Hogan, Inc., 699 F.2d 621, 625 (2d Cir.1983) (applying New York statute).

More important for our purposes than the continuing difficulty itself are its causes and its nature. And these plainly emerge in the cases. In the broadest sense, the cases demonstrate that once the dilution concept is sought to be given any form other than that of Schechter's simple original proposal it begins to lose its coherence as a legally enforceable norm. Specifically, it becomes difficult to identify the legal interest sought to be protected from "dilution," hence the legal harm sought to be prevented. As proposed by Schechter, the interest was easily identified as simply the mark's "uniqueness"--its singularity as a word-symbol contrived by its owner from outside "the human vocabulary"--and the harm, as a loss of that uniqueness. Schechter, Rational Basis of Trademark Protection, supra at 829. And, the legal cause of such a harm was the equally simple act of perfect or near-perfect replication of the senior mark by a junior mark. That all such "unique" marks had present or potential economic value--"selling power"--was assumed, as was the fact that any replication necessarily would "whittle away" that power. Id. at 830-31. Under that model, therefore, no proof would be required to prove dilution except the fact that a junior mark replicated the protected mark: no economic harm beyond that need be independently proved. See id. As commentators have fairly observed, the effect of this radical dilution model would have been to create property rights in gross in the narrow category of marks it protected, making them comparable (though without their time-limits) to those protected by patent and copyright law. See, e.g., Robert N. Klieger, Trademark Dilution: The Whittling Away of the Rational Basis for Trademark Protection, 58 U. Pitt. L. Rev. 789, 802 (1997) [hereinafter Klieger, Trademark Dilution ].

Although when the state antidilution statutes began to be enacted the dilution concept had found expression as a legal construct only in the form proposed by Schechter, it is clear that none of the original or following statutes purported to enact that specific form. Instead, without defining the term, their typical formulation simply proscribed and made subject to injunction the use of any mark that created a "[l]ikelihood of ... dilution of the distinctive quality of a [senior] mark ... notwithstanding the absence of competition between the parties or of confusion as to the source of goods or services." Model State Trademark Act § 12 (1964), reprinted in, J. Thomas McCarthy,3 McCarthy on Trademarks and Unfair Competition § 22:8 (4th ed.1998) [hereinafter 3 McCarthy ]. As earlier indicated, this bare-bones codification, centered on an unelaborated term of art having no previously acquired meaning through the common law decisional process, has puzzled courts from the outset as to just exactly what legal interest it sought to protect, and legal harm to prevent.3

While perfectly synthesizing the courts' varying approaches to the interpretive problem is impossible, a few observations relevant to our interpretive purpose can be ventured. The first is that though the typical state statute formulation is susceptible to an in-gross-property-right interpretation--by reading "distinctive quality" as essentially synonymous with "uniqueness" in the Schechter model--no court seems to have taken that blunt approach. Instead, frequently alluding to Schechter's identification of the senior mark's "selling power," and the "whittling away" of that power as the ultimate concerns of dilution's special protective function, the courts seem generally to have assumed that loss of that power, and the economic value it represents, was the end harm at which the antidilution statutes were aimed. See, e.g., Sally Gee, 699 F.2d at 624-25 ("The interest protected by [the New York antidilution statute] is ... the selling power that a distinctive mark or name with favorable associations has engendered for a product in the mind of the consuming public."); Polaroid Corp. v. Polaraid, Inc., 319 F.2d 830, 836 (7th Cir.1963) (" 'dilution is an infection which, if allowed to spread, will inevitably destroy the advertising value of the mark' ") (quoting Callmann, The Law of Unfair Competition and Trademarks 1643 (2d ed.1950)) (applying Illinois antidilution statute); The Munters Corp. v. Matsui Am., Inc., 730 F. Supp. 790, 802 (N.D.Ill.1989) (addressing whether defendant's use of its mark "will dilute the commercial magnetism of [plaintiff's] mark") (applying Illinois antidilution statute), aff'd, 909 F.2d 250 (1990).

The real interpretive problem has been with how harm to the senior mark's selling power resulting from the junior mark's use could be proved. Logic has compelled agreement that as a threshold matter some mental association of the two marks by a relevant universe of consumers must be proved (or presumed) in order to allow inference of the necessary causal connection between use and proven harm. See e.g., Mead Data, 875 F.2d at 1031 ("It is apparent ... that there must be some mental association between plaintiff's and defendant's marks.") (applying New York law); Fruit of the Loom, Inc. v. Girouard, 994 F.2d 1359, 1363 (9th Cir.1993) ("Whittling away will not occur unless there is at least some subliminal connection in a buyer's mind between the two parties' uses of their marks.") (applying California law). And, plain statutory text has directed that the threshold mental association and any resulting harm need only be proved as matters of future "likelihood." See, e.g., N.Y. Gen. Bus.Law § 360-1 ("Likelihood of ... dilution of the distinctive quality of a mark shall be a ground for injunctive relief...."); Cal. Bus. & Prof.Code § 14330 (same). But beyond these points of agreement the difficulty has remained: how, in the absence of any consumer confusion as to source, can harm to the senior mark's selling power traceable to the junior mark's use be proved even as likely future fact? On that, no consensus has emerged in judicial interpretations and applications of the state statutes. Three general approaches can be discerned in the efforts over time to deal with the problem.

Early on, during the period of general judicial hostility to the whole statutory dilution concept, see Restatement, supra § 25 cmt. b, some courts, seeming to assume that the requisite harm could only be shown by evidence of some form of product-diverting consumer confusion (presumably other than source confusion) invariably found such proof lacking. See, e.g., Cue Publishing Co. v. Colgate-Palmolive Co., 45 Misc. 2d 161, 168, 256 N.Y.S.2d 239, 245-46 (N.Y.Sup.Ct.),aff'd, 23 A.D.2d 829, 259 N.Y.S.2d 377 (1965).

Another approach, which assumed both that likelihood of harm to selling power must be proved, and that such harm could occur despite the absence of any consumer confusion, came in Judge Sweet's influential Mead Data proposal. Though not expressed in these exact procedural terms, the suggestion was effectively that proof must and could be sought through the normal judicial process of fact-finding by inference from a set of contextual factors (degree of mark and product similarity, etc.) that he proposed as relevant for the purpose. See Mead Data, 875 F.2d at 1035 (Sweet, J., concurring).

Finally, and most drastically, some courts have taken the position, in part at least for the very reason that they consider likelihood of harm to a mark's selling power to be incapable of direct or inferential proof, that it may simply be presumed from the identity or sufficient similarity of the marks. See Ringling Bros.-Barnum & Bailey Combined Shows, Inc. v. Celozzi-Ettelson Chevrolet, Inc., 855 F.2d 480, 484 (7th Cir.1988) (opining, in applying Illinois statute to uphold injunction against use of "Greatest Used Car Show on Earth" slogan, that "[w]ithout a likelihood of confusion there is no effective way to measure [Ringling's] loss of audience or potential growth," but presuming likelihood of that harm for injunctive purposes); Freedom Sav. & Loan Assoc. v. Way, 757 F.2d 1176, 1186 (11th Cir.1985) (opining, in applying Florida antidilution statute, that although "[d]ilution requires some proof that the use of a trademark decreases [a senior mark's] commercial value[,] [i]f the plaintiff holds a distinctive trademark, it is enough that the defendant has made significant use of a very similar mark"); see also Gaeta Cromwell, Inc. v. Banyan Lakes Village, 523 So. 2d 624, 626-27 (Fla.Dist.Ct.App.1988) (holding, under Florida statute, that where a senior mark "is strong and distinctive, mere 'significant use' of the name by the defendant is enough to establish decrease in its commercial value, and therefore 'dilution' "). As is obvious, by requiring no proof of "likely" dilution of a senior mark but the identity or sufficient similarity of a junior mark, this approach effectively interprets the state statutes as creating property rights in gross in the senior mark.4

From this fifty-year course of judicial experience, several propositions can be ventured. The first is that a general agreement has emerged that "dilution" under the state statutes involves as an essential element some form of harm to the protected mark's selling power--its economic value--resulting otherwise than by consumer confusion from the junior mark's use. Because the statutes require proof only of a "likelihood" of dilution, however, the courts have not been required to work out either the exact nature of the end economic harm it contemplates nor how, if at all, it could be proved as accomplished economic fact. Instead, looking only to the likelihood of such a speculative future condition, the courts have either (1) assumed that its essential elements--mental association, causation, harm--could be found (or rejected) as fact by inference from a balancing of the "Mead factors," or (2) assumed that all those elements could be conclusively presumed--direct or inferential proof being impossible--simply from proof of the identity or near-identity of the two marks.

From all this, it is evident that the most significant feature of the state antidilution statutes has been their requirement that only a "likelihood of dilution" rather than actual dilution need be proved to entitle a claimant to the injunctive relief which they provide as the sole statutory remedy. This has enabled the courts to avoid hard definition of the economic harm to a senior mark's "selling power" that they generally agree is an essential element of statutory "dilution." And, even more critically, the necessary speculativeness of any inquiry into future states and conditions has led some courts to allow the essential elements of "likely" dilution to be inferred as fact from the "Mead factors," or, even more drastically, to be presumed from no more than the identity or sufficient similarity of the two marks.

We have explored the judicial experience with state antidilution statutes at this length because of the needed light it sheds upon the significance of key contrasting provisions of the federal Act. And, because of the light shed in turn by those provisions upon the specific interpretive issue we consider: whether, as Ringling essentially contends, "dilution" under the federal Act requires no more proof than sufficient similarity of junior mark to senior to evoke in consumers an "instinctive mental association" of the two.

Two key provisions of the federal Act, considered in relation to the state statutes and their interpretation bear directly upon that issue and provide its answer. Most critically, the federal Act proscribes and provides remedy only for actual, consummated dilution and not for the mere "likelihood of dilution" proscribed by the state statutes. And, by specifically defining dilution as "the lessening of the capacity of a famous mark to identify and distinguish goods or services," the federal Act makes plain what the state statutes arguably may not: that the end harm at which it is aimed is a mark's selling power, not its "distinctiveness" as such.

Accepting these two critical points, we therefore interpret the Act, in general agreement with the district court, as requiring for proof of "dilution" (1) a sufficient similarity between the junior and senior marks to evoke an "instinctive mental association" of the two by a relevant universe of consumers which (2) is the effective cause of (3) an actual lessening of the senior mark's selling power, expressed as "its capacity to identify and distinguish goods or services."

This concededly is a stringent interpretation of "dilution" under the federal Act. It confines the federal dilution claim to a more narrow scope than that generally now accorded by courts to state-law dilution claims. But, given the critical provisions that expressly differentiate the federal Act on key points from the state statutes, we must assume that this was exactly what was intended by Congress. See I.P. Lund Trading ApS, Kroin Inc. v. Kohler Co., 163 F.3d 27, 45 (1st Cir.1998) (observing that "dilution" is a "term[ ] of art given specific, rigorous meaning[ ] by the [Act]"). It obviously is directly at odds with the mental-association-alone interpretation urged by Ringling, and Ringling challenges it in respects that require discussion.

Reflecting the unremitting difficulty of the interpretive problem for all who touch it, Ringling's challenge to the district court's interpretation is not a consistently clear one. In the end, however, it plainly comes to insistence that the federal Act does not require independent proof either that the senior mark has sustained any actual harm to its economic value or that use of the junior mark is the effective cause of any harm shown; instead, it is contended that the Act requires only proof of a sufficient visual similarity of the marks to evoke in consumers an "instinctive mental association" of the two.

To lay the basis for discussing that alternative and refuting interpretation, we note that in logic it could only be accepted (1) if the interest intended to be protected by the Act is the senior mark's "distinctiveness" as such, in the narrow sense of its singularity as a word symbol, or (2) if, although the Act requires proof of some actual harm to the senior mark's economic value caused by the junior mark's use, it permits both cause and harm to be judicially presumed as facts from the sufficient similarity of the two marks. As is evident, the first possibility describes the radical property-right-in-gross model of dilution as originally proposed by Schechter; the second describes the presumption process used by some courts to find a "likelihood of dilution" under state antidilution statutes. We are satisfied that the federal Act does not permit either of these interpretations.

Take first the property-right-in-gross interpretation. While Ringling does not press that interpretation in those exact terms, it is sufficiently implicit in its argument to require addressing.5 Doing so, we simply cannot believe that, as a general proposition, Congress could have intended, without making its intention to do so perfectly clear, to create property rights in gross, unlimited in time (via injunction), even in "famous" trademarks. Had that been the intention, it is one easily and simply expressed by merely proscribing use of any substantially replicating junior mark. And that surely is not what the Act says. However amorphously they may be expressed, and however difficult to prove in practice, the Act literally prescribes as elements of its dilution claim both specific harm to the senior mark's economic value in the form of a "lessening of [its] capacity ... to identify and distinguish goods and services," and a causal connection between that harm and the "commercial use" of a replicating junior mark. 15 U.S.C. §§ 1125(c)(1); 1127. It will not bear a property-right-in-gross interpretation.

Neither can the Act be interpreted to require proof of actual economic harm and its effective cause but permit them to be judicially presumed from proof alone of the marks' sufficient similarity. As earlier noted, that process has been used by some courts in applying state antidilution statutes that require proof only of a "likelihood of dilution." Whatever may be the justification for using it in that setting, it could not properly be used under a statute requiring proof of actual harm already caused by use of a junior mark. Under basic evidentiary presumption principles, the probabilities are not high enough nor means of proof sufficiently lacking to allow such a presumption. See 2 McCormick, supra § 343, at 454-55 (identifying probability estimates and proof difficulties as usual basis for judicial presumptions, with probability the "most important consideration").

Take first causation. As the district court in this case rightly observed, "marks can lose their distinctiveness or power to identify goods and services for various reasons other than the use of a junior mark." 955 F. Supp. at 615. And, for that reason, the court opined that junior mark use could not be judicially presumed to be the cause of any actual economic harm to the senior mark that might be proved. See id. We agree.

Nor, for similar reasons could actual harm itself be presumed. That economic harm inevitably will result from any replicating junior use is by no means that certain. See Klieger, Trademark Dilution, supra at 813 n. 134 ("[T]o say that the senior mark will inevitably lose its 'commercial magnetism' or 'selling power' is far from intuitive"). It is not at all improbable that some junior uses will have no effect at all upon a senior mark's economic value, whether for lack of exposure, general consumer disinterest in both marks' products, or other reasons. Indeed, common sense suggests that an occasional replicating use might even enhance a senior mark's "magnetism"--by drawing renewed attention to it as a mark of unshakable eminence worthy of emulation by an unthreatening non-competitor. Imitation, that is, may occasionally operate in the marketplace as in social manners as the "sincerest form of flattery." In any event, there are enough reasons why replicating junior use of a mark might not cause any actual economic harm to a senior mark that it is not a proper subject for judicial presumption.

Nor can it be said (as perhaps it can with respect to proving mere "likelihood of dilution") that even if it is the fact that actual economic harm caused by replicating junior use has occurred, there is no way to prove those facts independently. Though proof of those elements of the elusive dilution concept may tax the skills of advocacy, that results more from their substantive uncertainty than from lack of available means of proof. As we will discuss, if they do exist, there are means of proving them by normal evidentiary processes. Impossibility or near-impossibility of proving them does not support their judicial presumption.

Perhaps recognizing the difficulty of interpreting the federal Act in either of these ways, both of which assume that some actual harm must be proved but say either that the harm is only to a famous mark's distinctiveness as such or that, if economic harm is required, it may be judicially presumed from similarity of marks, Ringling's principal argument seems to be that if the Act requires proof of any form of economic harm, it is only threatened harm, not actual consummated harm. In short, Ringling argues that though the Act does not literally proscribe mere "likelihood of dilution" in the manner of state antidilution statutes, that is its intended meaning. And, from that the argument implicitly runs that merely future harm could be much more easily proved (or judicially presumed?) than can the actual, consummated economic harm that the district court's interpretation requires be independently proved.

The literal parsing argument for mere threatened, future economic harm focuses on the word "capacity" in the Act's definition of "dilution" as the "lessening of the capacity of a famous mark to identify and distinguish goods or services." 15 U.S.C. § 1127. "Capacity," the argument runs, necessarily imports futurity; it means in this context the ability of a mark continuously over future time to "identify and distinguish," even if it has not yet suffered any lessening of that ability. We cannot accept that interpretation as a matter of the Act's plain meaning. We think it is belied both by the word's ordinary intrinsic meaning and by its use in context. One can surely speak expressly of "present capacity," or of "future capacity," or of a "former capacity," but unless it is so temporally modified or otherwise given its intended temporal meaning, the word is neutral in that respect. In context here, it is plain that the "capacity" spoken of is "former capacity." The verb of which it is the object is the clear indicator; the conduct proscribed is that which "lessens" capacity, not that which "will" or "may"lessen. Id. There are other contextual indicators. The conduct proscribed is "another person's ... use," not merely threatened use; that "causes," not that "will" or "may" cause. Id. § 1125(c)(1). Unlike the state antidilution statutes which provide only injunctive relief, reflecting their sole focus on prevention of future harm, the federal Act provides that where willful conduct is shown, both compensatory and restitutionary relief may be awarded--for necessarily consummated economic harm. Id. §§ 1125(c)(2), 1117(a), 1118. Finally and most telling, there is the fact that in the face of the obvious centrality of "likelihood of dilution" provisions in the interpretation and application of state antidilution statutes for the fifty years of their existence,the federal Act does not so provide.6

For all these reasons, we agree with the district court that to establish dilution of a famous mark under the federal Act requires proof that (1) a defendant has made use of a junior mark sufficiently similar to the famous mark to evoke in a relevant universe of consumers a mental association of the two that (2) has caused (3) actual economic harm to the famous mark's economic value by lessening its former selling power as an advertising agent for its goods or services.

We turn now to Ringling's challenges to the district court's determination that on the evidence adduced, Ringling had not proved dilution under the federal Act. We review that determination, as one of mixed law and fact, under the clearly erroneous standard of Fed. R.Civ. P. 52(a). See Miller v. Mercy Hosp., Inc., 720 F.2d 356, 361 (4th Cir.1983) (reversal is warranted under the clearly erroneous standard when "the findings under review were induced by an erroneous view of the controlling legal standard, or are not supported by substantial evidence, or were made without properly taking into account substantial evidence to the contrary or are against the clear weight of the evidence considered as a whole") (citations omitted); cf. Jordache Enters. Inc. v. Hogg Wyld, Ltd., 828 F.2d 1482, 1489 (10th Cir.1987) (holding, in applying New Mexico antidilution statute, that "[d]ilution ... is a question of fact ... review[ed] under the clearly erroneous standard").

As indicated, that evidence consisted of the general background facts summarized in Part I of this opinion and the results of a consumer survey introduced by Ringling. The district court of course assessed the evidence under its interpretation of the Act, and because we agree with that interpretation, we review its findings against that legal standard. And, because the court looked separately at the consumer survey evidence and the "Mead " factors in assessing the overall weight of the evidence, we proceed in the same way.

It is important to remember that in keeping with its legal theory that it need prove only an "instinctive mental association" of the two marks, Ringling's survey was designed to develop only that fact. And, correspondingly, that in assessing the evidence under its quite different interpretation, the district court was looking for more: for actual harm to the senior mark's capacity to "identify and distinguish" resulting from any mental association of the marks evoked for consumers by the junior mark's use.

We begin our review of the district court's assessment of the survey evidence by summarizing the survey's methodology and results. The survey was conducted by interviewing individuals at seven shopping malls throughout the country, including one in Utah. At each location, randomly selected shoppers were presented with a card containing the fill-in-the-blank statement "THE GREATEST _____ ON EARTH" and were asked what word or words they would use to complete the phrase. If the shoppers completed the statement, they were asked with whom or what they associated the completed statement. And, they were asked further whether they could think of any other way to complete the statement, and with whom or what they associated the resulting statement.

The survey results showed that in Utah (1) 25% of the respondents completed the statement THE GREATEST _____ ON EARTH with only the word "show" and associated the completed statement with the Circus; (2) 24% completed the statement with only the word "snow" and associated the completed statement with Utah; and (3)21% of respondents completed the statement with "show" and associated the result with the Circus and also completed the statement with "snow" and associated the completed statement with Utah. The survey further showed that outside of Utah (1) 41% of respondents completed the statement THE GREATEST _____ ON EARTH with only the word "show" and associated the completed statement with the Circus; (2) 0% completed the statement with only the word "snow" and associated the completed statement with Utah; and (3) fewer than 0.5% of respondents completed the statement with "show" and associated the result with the Circus and also completed the statement with "snow" and associated the completed statement with Utah. (Id. at 484.)

The district court concluded that this evidence failed to show dilution under the Act. In the first place, the court found the survey results inadequate to prove that consumers even made the requisite threshold mental association of the marks. When faced with the fill-in-the-blank phrase, "The Greatest _____ on Earth," some consumers filled in the blank with both the words "show" and "snow." When asked with whom or what they associated the completed phrase "the Greatest Show on Earth," virtually every consumer--inside Utah and outside Utah--indicated in one way or other that they only associated Ringling's circus with that phrase. When asked with whom or what they associated the completed phrase, "The Greatest Snow on Earth," every single consumer--inside Utah and outside Utah--indicated that they only associated Utah with the completed phrase. Not one consumer indicated that he associated the phrase, "The Greatest Show on Earth" with the phrase, "The Greatest Snow on Earth." Summarizing these results, the district court concluded that they were "strong evidence of the absence of dilution, not the presence of it," 955 F. Supp. at 617, that is, that they tended to disprove rather than to prove the required threshold mental association of the marks.

And the court found the survey results even more lacking as proof that Utah's use of its mark had caused any "lessening" of the "capacity" of Ringling's trademark slogan to "identify and distinguish" its circus as the mark's subject. Specifically, the court pointed to survey results indicating that consumer familiarity with Ringling's mark was greater in Utah (46%), where Utah's mark was well-known, than in the rest of the country (41%), where Utah's mark was virtually unknown, and that virtually every viewer questioned associated Ringling's mark (as distinguished from the fill-in-the-blank slogan) with, and only with, the Ringling circus and not "with Utah, with wintersports, or with any activities that are attributable to Utah's use of [its mark]." Id. at 618.

We affirm as not clearly erroneous the district court's assessment that the consumer survey evidence does not support a finding of dilution under the federal Act. While we might have some concern with the implicit finding that this evidence does not even show the requisite threshold "mental association" of the two marks within the consumer market surveyed, we have no concern respecting the specific finding that the survey evidence does not show that use of Utah's junior mark had caused any actual harm to Ringling's mark in the form of a lessening of that mark's former capacity to identify and distinguish Ringling's circus as its subject. And that, of course, suffices to support the court's ultimate conclusion.

Ringling makes several challenges to this assessment of its survey evidence. We address only its principal contention: that the court's assessment was flawed by a legal misapprehension that the "mental association" of marks requisite to proof of dilution must go beyond mere association of the marks in isolation and involve some mistake or confusion as to the marks and their respective goods or services. This was indeed the court's position, and it of course traces to the court's correct interpretation of the Act which, at odds with Ringling's position, requires not just proof of visual similarity sufficient to evoke mental association of the marks, but actual harm resulting from that association to the senior mark's selling power. While we might not have expressed the broader "mental association" requirement in exactly those terms, we think it accurately captured the proof requirements imposed by that court's legal interpretation, which we have affirmed as the correct one. What the district court was saying in effect was this. If you seek to rely for proof of dilution only upon evidence of the mental impressions evoked in consumers upon viewing the marks, then those impressions must go beyond mere recognition of a visual similarity of the two marks to allow a reasonable inference that the junior mark's use has caused actual harm to the senior mark's selling or advertising power.

That accurately reflects the interpretation of the Act that we have affirmed and confirms that the court's assessment of the survey evidence under that legal interpretation must be affirmed.

We discuss only briefly the district court's alternative assessment of the evidence under the so-called "Mead factors." Recognizing that this judicial fact-finding process had been used by various courts as a device by which the "likelihood of dilution" might be inferred under state antidilution statutes from a set of relevant factors--similarity of marks, similarity of products, consumer sophistication, predatory intent, and renown of the two marks--the court, somewhat puzzlingly, undertook to apply it to this claim brought under the federal Act. Doing so on the basis of a run-through of the whole set and a final "balancing" of their separate indicators, the court concluded that neither did that evidence "demonstrate by a preponderance ... that Utah's mark lessens the capacity of [Ringling's mark] to identify and distinguish Ringling's circus." Id. at 622.

Ringling of course challenges the court's fact-finding by that process, suggesting that some of the factors are wholly inappropriate under the federal Act, and that if applied under that Act they require a different weighting than that accorded them by the district court. Utah of course defends the court's use of the device, but has to confess the inappropriateness of some of these factors developed under state antidilution statutes to cases under the federal Act.

We are persuaded, in agreement with other courts and commentators, and in probable agreement with some of the obvious discomfort of both parties in this case, that, by and large, the Mead-factor analysis simply is not appropriate for assessing a claim under the federal Act. See Lund, 163 F.3d at 49-50; see also Klieger, Trademark Dilution, supra at 826-27. As we have earlier noted, the process has obvious utility in making the long leaps of inference that can be used to find a mere "likelihood of dilution," but inferring actual harm and effective causation from such factors as "consumer sophistication,"and "predatory intent" is a chancy process at best. Indeed, of the factors, only mark similarity and, possibly, degree of "renown" of the senior mark would appear to have trustworthy relevance under the federal Act. See 3 McCarthy, supra § 24:94.1.

In consequence, we conclude that though the district court's use of this process as a complementary fact-finding process was inappropriate, its use can be treated as one having no ultimate prejudicial effect upon Ringling as appellant.

We conclude this part of our discussion with some observations about the problems associated with proving dilution under the federal Act interpreted, as we do, to require proof of actual dilution caused by junior mark use. We think this is directly relevant to the interpretive issue on which this case turns.

Courts must of course presume in interpreting statutes creating new civil causes of action that they have enforceable substantive content. So, within permissible interpretive bounds, they must seek to find such a meaning rather than ascribe to Congress the intended or inadvertent doing of a vain legislative deed. One reason to shy away from a possible interpretation--indeed, one reason to find it not possibly the intended legislative meaning--is that it defies proof by the ordinary processes of advocacy. That interpretive phenomenon is apparent in the early reactions of courts and commentators to the federal Act's dilution concept.

The difficulties of proving actual dilution by practically available means is evident--as we have recognized, at length. It may have led a few federal courts early on simply to assume, without facing up to the interpretive difficulty of doing so, that the federal Act only requires proof of a "likelihood of dilution." See Elvis Presley Enters., Inc. v. Capece, 950 F. Supp. 783, 797 (S.D.Tex.1996), rev'd on other grounds, 141 F.3d 188 (5th Cir.1998); Ringling Bros.-Barnum & Bailey Combined Shows, Inc. v. B.E. Windows Corp., 937 F. Supp. 204, 209, 211 (S.D.N.Y.1996); Hartz & Co., Inc. v. Italia, Inc., No.97 Civ. 5657(RO), 1998 WL 132787, at * 1 & n. 9 (S.D.N.Y. Mar. 24, 1998). Going this route permits proof of federal Act dilution by the long leaps of inference to, and judicial presumption of, future harm that have been considered allowable by some courts applying state antidilution statutes. And, it avoids the problems of proving actual, consummated harm effectively caused by junior mark use.

On the other hand, the perceived difficulties could lead to an interpretation that effectively requires proof of actual harm only to the mark's distinctiveness as such. This in fact is what the dilution-policy critic, Klieger, suggests must have been the intended meaning--though it was not the plain meaning--of the federal statute as enacted. See Klieger, Trademark Dilution, supra at 840-41.

We think that proof of actual dilution cannot be considered impossible and therefore not possibly what Congress could have intended. Proof will be difficult, because actual, consummated dilutive harm and its cause are difficult concepts. But the concept is a substantively viable one, and the means of proof are available.

It surely must be possible in the marketing world that the replicating use of a "famous" mark causes it actually to lose to some extent the "selling power" deriving from its distinctiveness that it formerly had. And, without attempting to chart the exact shape and course of advocacy that might prove that occurrence, we think three general means are available. Most obviously, but most rarely, there might be proof of an actual loss of revenues, and proof of replicating use as cause by disproving other possible causes. Most obviously relevant, and readily available, is the skillfully constructed consumer survey designed not just to demonstrate "mental association" of the marks in isolation, but further consumer impressions from which actual harm and cause might rationally be inferred. See Bible, Defining and Quantifying Dilution, supra at 327-28 (proposing a survey designed to prove actual dilution and noting: "An effective survey ... must establish not only that consumers associate the mark with both parties, but also that some quanta of the original mark's identifying ability or selling power has been diminished."). Finally, relevant contextual factors such as the extent of the junior mark's exposure, the similarity of the marks, the firmness of the senior mark's hold, are of obvious relevance as indirect evidence that might complement other proof.

We turn finally to Ringling's challenge to the district court's order striking its demand for a jury trial. The district court first concluded that the Act did not provide a right to jury trial. See Ringling Bros.-Barnum & Bailey, Combined Shows, Inc. v. Utah Div. of Travel Dev., 955 F. Supp. 598, 599-601 (E.D.Va.1997). Turning to whether the Seventh Amendment, however, compels the right the court noted that except where a defendant "willfully intended ... to cause dilution," the only remedy available to a dilution plaintiff is an injunction. See id. at 603 (quoting 15 U.S.C. § 1125(c)(2)). And, it then properly held that "where only an injunction is available to remedy dilution, the Seventh Amendment does not compel a jury trial." Id.; see generally 9 Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure § 2308 (2d ed.1994) ("[T]here is no constitutional right to a jury trial on a claim for an injunction."). And, then concluding that Utah was entitled to summary judgment on the issue of "willful intent," the district court ruled that because Ringling was therefore limited to injunctive relief on its dilution claim, it was not constitutionally entitled to a jury trial. See 955 F. Supp. at 603. "Alternatively," the district court held, "even if 'willful intent' is shown, the Seventh Amendment does not compel a jury trial for Ringling's dilution claim because there is no evidence of damages and the remaining remedies in §§ 1117(a) and 1118 are wholly equitable." Id.7

Ringling argues that the district court erred in granting summary judgment on the issue of willful intent. And, conceding as it must that it failed to introduce any evidence of actual damages, Ringling argues that this should not disentitle it to jury trial because a "reasonable royalty" should be awarded as monetary relief in lieu of actual damages. (Appellant's Br. at 32-39.)8

We review the district court's summary judgment ruling de novo. See Egbuna v. Time-Life Libraries, Inc., 153 F.3d 184, 186 (4th Cir.1998). Because we conclude that the district court did not err in granting Utah's motion for summary judgment on the issue of willful intent, we need not address whether it might alternatively have been granted because of the lack of any evidence of actual damages. And, because even if there were a constitutional right to jury trial on the willfulness issue, it would not be infringed by a proper disposition of the issue by summary judgment, see Moore's Federal Practice, § 38.12[a][i], we may reserve decision on the constitutional question for another day. See Ashwander v. Tenn. Valley Auth., 297 U.S. 288, 347, 56 S. Ct. 466, 80 L. Ed. 688 (1935) (Brandeis, J., concurring.).

Ringling contends that the district court "embraced an incorrect legal standard for proving willfulness," erroneously requiring it to show that Utah had developed a willful intent when it adopted its mark. (Appellant's Br. at 37.) According to Ringling, the Act "places no temporal limitation on when the junior mark owner develops an intent, but only that the junior user does develop such an intent." (Id. at 37 n. 15 (emphasis omitted).) This argument, even if successful,would not carry the day for Ringling for even if all of Ringling's evidence of willful intent, without temporal limitation is considered, it does not suffice to raise a genuine issue of material fact.

Ringling proffers two pieces of evidence on the issue. First, it proffers evidence that Utah produced promotional merchandise for the 2002 winter Olympics that included the Utah Olympic symbol and the GREATEST SNOW mark. While there can be no willful intent to dilute without use of the allegedly diluting mark, that standing alone does not suffice. The second item of evidence, building on the first, is that Utah had access to the GREATEST SHOW mark prior to the time Utah adopted its mark and that the marks are strikingly similar. As the district court noted, however, Utah's adoption and use of its mark notwithstanding its knowledge of Ringling's mark, is "as consistent with a belief that Utah's mark does not dilute as ... with an allegation of 'willful intent.' " See 955 F. Supp. at 603-04. And, as the district court noted, the record reflects that Utah had a good faith, reasonable belief that its use of the GREATEST SNOW mark was lawful. See id. at 604 & n. 16. The two items of evidence do not suffice when considered together to create a genuine issue of material fact precluding summary judgment.

For these reasons we conclude that the district court properly granted Utah's motion for summary judgment on the issue of willfulness and on that basis properly struck Ringling's demand for a jury trial.

AFFIRMED.