Mr. Justice Jackson

delivered the opinion of the Court.

Linde Air Products Co., owner of the Jones patent for an electric welding process and for fluxes to be used therewith, brought an action for infringement against Lincoln and the two Graver companies. The trial court held four flux claims valid and infringed and certain other flux claims and all process claims invalid. 75 U. S. P. Q. 231. The Court of Appeals affirmed findings of validity and infringement as to the four flux claims but reversed the trial court and held valid the process claims and the remaining contested flux claims. 167 F. 2d 531. We granted certiorari, 335 U. S. 810, and reversed the judgment of the Court of Appeals insofar as it reversed that of the trial court, and reinstated the District Court decree. 336 U. S. 271. Rehearing was granted, limited to the question of infringement of the four valid flux claims and to the applicability of the doctrine of equivalents to findings of fact in this case. 337 U. S. 910.

*607At the outset it should be noted that the single issue before us is whether the trial court’s holding that the four flux claims have been infringed will be sustained. Any issue as to the validity of these claims was unanimously determined by the previous decision in this Court and attack on their validity cannot be renewed now by reason of limitation on grant of rehearing. The disclosure, the claims, and the prior art have been adequately described in our former opinion and in the opinions of the courts below.

In determining whether an accused device or composition infringes a valid patent, resort must be had in the first instance to the words of the claim. If accused matter falls clearly within the claim, infringement is made out and that is the end of it.

But courts have also recognized that to permit imitation of a patented invention which does not copy every literal detail would be to convert the protection of the patent grant into a hollow and useless thing. Such a limitation would leave room for — indeed encourage — the unscrupulous copyist to make unimportant and insubstantial changes and substitutions in the patent which, though adding nothing, would be enough to take the copied matter outside the claim, and hence outside the reach of law. One who seeks to pirate an invention, like one who seeks to pirate a copyrighted book or play, may be expected to introduce minor variations to conceal and shelter the piracy. Outright and forthright duplication is a dull and very rare type of infringement. To prohibit no other would place the inventor at the mercy of verbalism and would be subordinating substance to form. It would deprive him of the benefit of his invention and would foster concealment rather than disclosure of inventions, which is one of the primary purposes of the patent system.

*608The doctrine of equivalents evolved in response to this experience. The essence of the doctrine is that one may not practice a fraud on a patent. Originating almost a century ago in the case of Winans v. Denmead, 15 How. 330, it has been consistently applied by this Court and the lower federal courts, and continues today ready and available for utilization when the proper circumstances for its application arise. “To temper unsparing logic and prevent an infringer from stealing the benefit of an invention”1 a patentee may invoke this doctrine to proceed against the producer of a device “if it performs substantially the same function in substantially the same way to obtain the same result.” Sanitary Refrigerator Co. v. Winters, 280 U. S. 30, 42. The theory on which it is founded is that “if two devices do the same work in substantially the same way, and accomplish substantially the same result, they are the same, even though they differ in name, form, or shape.” Machine Co. v. Murphy, 97 U. S. 120, 125. The doctrine operates not only in favor of the patentee of a pioneer or primary invention, but also for the patentee of a secondary invention consisting of a combination of old ingredients which produce new and useful results, Imhaeuser v. Buerk, 101 U. S. 647, 655, although the area of equivalence may vary under the circumstances. See Continental Paper Bag Co. v. Eastern Paper Bag Co., 210 U. S. 405, 414-415, and cases cited; Seymour v. Osborne, 11 Wall. 516, 556; Gould v. Rees, 15 Wall. 187, 192. The wholesome realism of this doctrine is not always applied in favor of a patentee but is sometimes used against him. Thus, where a device is so far changed in principle from a patented article that it performs the same or a similar function in a substantially different way, but nevertheless falls within the *609literal words of the claim, the doctrine of equivalents may be used to restrict the claim and defeat the patentee’s action for infringement. Westinghouse v. Boyden Power Brake Co., 170 U. S. 537, 568. In its early development, the doctrine was usually applied in cases involving devices where there was equivalence in mechanical components. Subsequently, however, the same principles were also applied to compositions, where there was equivalence between chemical ingredients. Today the doctrine is applied to mechanical or chemical equivalents in compositions or devices. See discussions and cases collected in 3 Walker on Patents (Deller’s ed. 1937) §§ 489-492; Ellis, Patent Claims (1949) §§ 59-60.

What constitutes equivalency must be determined against the context of the patent, the prior art, and the particular circumstances of the ease. Equivalence, in the patent law, is not the prisoner of a formula and is not an absolute to be considered in a vacuum. It does not require complete identity for every purpose and in every respect. In determining equivalents, things equal to the same thing may not be equal to each other and, by the same token, things for most purposes different may sometimes be equivalents. Consideration must be given to the purpose for which an ingredient is used in a patent, the qualities it has when combined with the other ingredients, and the function which it is intended to perform. An important factor is whether persons reasonably skilled in the art would have known of the interchangeability of an ingredient not contained in the patent with one that was.

A finding of equivalence is a determination of fact. Proof can be made in any form: through testimony of experts or others versed in the technology; by documents, including texts and treatises; and, of course, by the disclosures of the prior art. Like any other issue of fact, final determination requires a balancing of credibility, *610persuasiveness and weight of evidence. It is to be decided by the trial court and that court’s decision, under general principles of appellate review, should not be disturbed unless clearly erroneous. Particularly is this so in a field where so much depends upon familiarity with specific scientific problems and principles not usually contained in the general storehouse of knowledge and experience.

In the case before us, we have two electric welding compositions or fluxes: the patented composition, Union-melt Grade 20, and the accused composition, Lincolnweld 660. The patent under which Unionmelt is made claims essentially a combination of alkaline earth metal silicate and calcium fluoride; Unionmelt actually contains, however, silicates of calcium and magnesium, two alkaline earth metal silicates. Lincolnweld’s composition is similar to Unionmelt’s, except that it substitutes silicates of calcium and manganese — the latter not an alkaline earth metal — for silicates of calcium and magnesium. In all other respects, the two compositions are alike. The mechanical methods in which these compositions are employed are similar. They are identical in operation and produce the same kind and quality of weld.

The question which thus emerges is whether the substitution of the manganese which is not an alkaline earth metal for the magnesium which is, under the circumstances of this case, and in view of the technology and the prior art, is a change of such substance as to make the doctrine of equivalents inapplicable; or conversely, whether under the circumstances the change was so insubstantial that the trial court’s invocation of the doctrine of equivalents was justified.

Without attempting to be all-inclusive, we note the following evidence in the record: Chemists familiar with the two fluxes testified that manganese and magnesium were similar in many of their reactions (R. 287, 669). There is testimony by a metallurgist that alkaline earth *611metals are often found in manganese ores in their natural state and that they serve the same purpose in the fluxes (R. 831-832); and a chemist testified that “in the sense of the patent” manganese could be included as an alkaline earth metal (R. 297). Much of this testimony was corroborated by reference to recognized texts on inorganic chemistry (R. 332). Particularly important, in addition, were the disclosures of the prior art, also contained in the record. The Miller patent, No. 1,754,566, which preceded the patent in suit, taught the use of manganese silicate in welding fluxes (R. 969, 971). Manganese was similarly disclosed in the Armor patent, No. 1,467,825, which also described a welding composition (R. 1346). And the record contains no evidence of any kind to show that Lincolnweld was developed as the result of independent research or experiments.

It is not for this Court to even essay an independent evaluation of this evidence. This is the function of the trial court. And, as we have heretofore observed, “To no type of case is this . . . more appropriately applicable than to the one before us, where the evidence is largely the testimony of experts as to which a trial court may be enlightened by scientific demonstrations. This trial occupied some three weeks, during which, as the record shows, the trial judge visited laboratories with counsel and experts to observe actual demonstrations of welding as taught by the patent and of the welding accused of infringing it, and of various stages of the prior art. He viewed motion pictures of various welding operations and tests and heard many experts and other witnesses.” 336 U. S. 271, 274-275.

The trial judge found on the evidence before him that the Lincolnweld flux and the composition of the patent in suit are substantially identical in operation and in result. He found also that Lincolnweld is in all respects equivalent to Unionmelt for welding purposes. And he concluded that “for all practical purposes, manganese silicate *612can be efficiently and effectually substituted for calcium and magnesium silicates as the major constituent of the welding composition.” These conclusions are adequately supported by the record; certainly they are not clearly erroneous.2

It is difficult to conceive of a case more appropriate for application of the doctrine of equivalents. The disclosures of the prior art made clear that manganese silicate was a useful ingredient in welding compositions. Specialists familiar with the problems of welding compositions understood that manganese was equivalent to and could be substituted for magnesium in the composition of the patented flux and their observations were confirmed by the literature of chemistry. Without some explanation or indication that Lincolnweld was developed by independent research, the trial court could properly infer that the accused flux is the result of imitation rather than experimentation or invention. Though infringement was not literal, the changes which avoid literal infringement are colorable only. We conclude that the trial court’s judgment of infringement respecting the four flux claims was proper, and we adhere to our prior decision' on this aspect of the case.

Affirmed.

Mr. Justice Minton took no part in the consideration or decision of this case.

Mr. Justice Black,

with whom

Mr. Justice Douglas

concurs, dissenting.

1 heartily agree with the Court that “fraud” is bad, “piracy” is evil, and “stealing” is reprehensible. But in *613this case, where petitioners are not charged with any such malevolence, these lofty principles do not justify the Court’s sterilization of Acts of Congress and prior decisions, none of which are even mentioned in today’s opinion.

The only patent claims involved here describe respondent’s product as a flux “containing a major proportion of alkaline earth metal silicate.” The trial court found that petitioners used a flux “composed principally of manganese silicate.” Finding also that “manganese is not an alkaline earth metal,” the trial court admitted that petitioners’ flux did not “literally infringe” respondent’s patent. Nevertheless it invoked the judicial “doctrine of equivalents” to broaden the claim for “alkaline earth metals” so as to embrace “manganese.” On the ground that “the fact that manganese is a proper substitute . . . is fully disclosed in the specification” of respondent’s patent, it concluded that “no determination need be made whether it is a known chemical fact outside the teachings of the patent that manganese is an equivalent . . . .” Since today’s affirmance unquestioningly follows the findings of the trial court, this Court necessarily relies on what the specifications revealed.1 In so doing, it violates a direct mandate of Congress without even discussing that mandate.

R. S. § 4888, as amended, 35 U. S. C. § 33, provides that an applicant “shall particularly point out and distinctly claim the part, improvement, or combination which he claims as his invention or discovery.” We have held in this very case that this statute precludes invoking the specifications to alter a claim free from ambiguous language, since “it is the claim which measures the grant *614to the patentee.” 2 Graver Mfg. Co. v. Linde Co., 336 U. S. 271, 277. What is not specifically claimed is dedicated to the public. See, e. g., Miller v. Brass Co., 104 U. S. 350, 352. For the function of claims under R. S. § 4888, as we have frequently reiterated, is to exclude from the patent monopoly field all that is not specifically claimed, whatever may appear in the specifications. See, e. g., Marconi Wireless Co. v. United States, 320 U. S. 1, 23, and cases there cited. Today the Court tacitly rejects those cases. It departs from the underlying principle which, as the Court pointed out in White v. Dunbar, 119 U. S. 47, 51, forbids treating a patent claim “like a nose of wax which may be turned and twisted in any direction, by merely referring to the specification, so as to make it include something more than, or something different from, what its words express. . . . The claim is a statutory requirement, prescribed for the very purpose of making the patentee define precisely what his invention is; and it is unjust to the public, as well as an evasion of the law, to construe it in a manner different from the plain import of its terms.” Giving this patentee the benefit of a grant that it did not precisely claim is no less “unjust to the public” and no less an evasion of R. S. § 4888 merely because done in the name of the “doctrine of equivalents.”

In seeking to justify its emasculation of R. S. § 4888 by parading potential hardships which literal enforcement might conceivably impose on patentees who had for some reason failed to claim complete protection for their discoveries, the Court fails even to mention the program for alleviation of such hardships which Congress itself *615has provided. 35 U. S. C. § 64 authorizes reissue of patents where a patent is “wholly or partly inoperative” due to certain errors arising from “inadvertence, accident, or mistake” of the patentee. And while the section does not expressly permit a patentee to expand his claim, this Court has reluctantly interpreted it to justify doing so. Miller v. Brass Co., 104 U. S. 350, 353-354. That interpretation, however, was accompanied by a warning that “Reissues for the enlargement of claims should be the exception and not the rule.” Id. at 355. And Congress was careful to hedge the privilege of reissue by exacting conditions. It also entrusted the Patent Office, not the courts, with initial authority to determine whether expansion of a claim was justified,3 ****8 and barred suits for retroactive infringement based on such expansion. Like the Court’s opinion, this congressional plan adequately protects patentees from “fraud,” “piracy,” and “stealing.” Unlike the Court's opinion, it also protects businessmen from retroactive infringement suits and judicial expansion of a monopoly sphere beyond that which a patent expressly authorizes. The plan is just, fair, and reasonable. In effect it is nullified by this decision undercutting what *616the Court has heretofore recognized as wise safeguards. See Milcor Steel Co. v. Fuller Co., 316 U. S. 143, 148. One need not be a prophet to suggest that today’s rhapsody on the virtue of the “doctrine of equivalents” will, in direct contravention of the Miller case, supra, make enlargement of patent claims the “rule” rather than the “exception.”

Whatever the merits of the “doctrine of equivalents” where differences between the claims of a patent and the allegedly infringing product are de minimis, colorable only, and without substance, that doctrine should have no application to the facts of this case. For the differences between respondent’s welding substance and petitioners’ claimed flux were not nearly so slight. The claims relied upon here did not involve any mechanical structure or process where invention lay in the construction or method rather than in the materials used. Rather they were based wholly on using particular materials for a particular purpose. Respondent’s assignors experimented with several metallic silicates, including that of manganese. According to the specifications (if these are to be considered) they concluded that while several were “more or less efficacious in our process, we prefer to use silicates of the alkaline earth metals.” Several of their claims which this Court found too broad to be valid encompassed manganese silicate; the only claims found valid did not. Yet today the Court disregards that crucial deficiency, holding those claims infringed by a composition of which 88.49% by weight is manganese silicate.

In view of the intense study and experimentation of respondent’s assignors with manganese silicate, it would be frivolous to contend that failure specifically to include that substance in a precise claim was unintentional. Nor does respondent attempt to give that or any other explanation for its omission. But the similar use of manganese in prior expired patents, referred to in the Court’s opinion, raises far more than a suspicion that its elimina*617tion from the valid claims stemmed from fear that its inclusion by name might result in denial or subsequent invalidation’ of respondent’s patent.

Under these circumstances I think petitioners had a right to act on the belief that this Court would follow the plain mandates of Congress that a patent’s precise claims mark its monopoly boundaries, and that expansion of those claims to include manganese could be obtained only in a statutory reissue proceeding. The Court’s ruling today sets the stage for more patent “fraud” and “piracy” against business than could be expected from faithful observance of the congressionally enacted plan to protect business against judicial expansion of precise patent claims. Hereafter a manufacturer cannot rely on what the language of a patent claims. He must be able, at the peril of heavy infringement damages, to forecast how far a court relatively unversed in a particular technological field will expand the claim’s language after considering the testimony of technical experts in that field. To burden business enterprise on the assumption that men possess such a prescience bodes ill for the kind of competitive economy that is our professed goal.

The way specific problems are approached naturally has much to do with the decisions reached. A host of prior cases, to some of which I have referred, have treated the 17-year monopoly authorized by valid patents as a narrow exception to our competitive enterprise system. For that reason, they have emphasized the importance of leaving business men free to utilize all knowledge not preempted by the precise language of a patent claim. E. g., Sontag Stores Co. v. Nut Co., 310 U. S. 281, and cases there cited. In the Sontag case Mr. Justice McReynolds, speaking for a unanimous Court, said in part: “In the case under consideration the patentee might have included in the application for the original patent, claims broad enough to embrace petitioner’s accused machine, but did not. *618This 'gave the public to understand’ that whatever was not claimed 'did not come within his patent and might rightfully be made by anyone.’ ” Id. at 293.

The Court’s contrary approach today causes it to retreat from this sound principle. The damages retroactively assessed against petitioners for what was authorized until today are but the initial installment on the cost of that retreat.

Mr. Justice Douglas,

dissenting.

The Court applies the doctrine of equivalents in a way which subverts the constitutional and statutory scheme for the grant and use of patents.

The claims of the patent are limited to a flux “containing a major proportion of alkaline earth metal silicate.” Manganese silicate, the flux which is held to infringe, is not an alkaline earth metal silicate. It was disclosed in the application and then excluded from the claims. It therefore became public property. See Mahn v. Harwood, 112 U. S. 354, 361. It was, to be sure, mentioned in the specifications. But the measure of the grant is to be found in the claims, not in the specifications. Milcor Steel Co. v. Fuller Co., 316 U. S. 143, 145, 146. The specifications can be used to limit but never to expand the claim. See McClain v. Ortmayer, 141 U. S. 419, 424.

The Court now allows the doctrine of equivalents to erase those time-honored rules. Moreover, a doctrine which is said to protect against practicing “a fraud on a patent” is used to extend a patent to a composition which could not be patented. For manganese silicate had been covered by prior patents, now expired. Thus we end with a strange anomaly: a monopoly is obtained on an unpatented and unpatentable article.

BOUDIN, Circuit Judge.

Before us are cross-appeals in a decade-long copyright dispute concerning two sitcoms on Puerto Rican television. On one side is TMTV, Corp., a successor in interest to the first sitcom’s production company; on the other, the second sitcom’s production company, Mass Productions, Inc., and its principals — including the star of both shows, Emmanuel “Sunshine” Logroño. The district court granted summary judgment to TMTV on its claim charging infringement; and a jury granted it substantial damages, which the district court reduced by an amount recovered by TMTV in settlement of related litigation.

*467A summary of events and prior proceedings will suffice. In 1997, Antonio Mojena- — an established television producer in Puerto Rico — approached two veteran entertainers — Sunshine Logroño and Iris Chacón — about co-hosting a variety show on Puerto Rican television. The show would be called De Noche con Iris y Sunshine and would begin with four two-hour episodes, continuing thereafter if it proved popular. Chacón and Logroño both agreed to participate.

During Mojena’s initial pitch, he and Logroño discussed filling some of the two-hour episodes of De Noche con Iris y Sunshine with comedy segments. To that end, Logroño arranged a meeting in late September or early October 1997 with a team of comedians and scriptwriters, including scriptwriters Roberto Jiménez and Miguel Morales; Jiménez had worked with Logroño before, while Morales had not. At the meeting Jiménez suggested a sitcom centered on a condominium building and its residents. Logroño approved of the idea of setting a comedy segment in a condominium building, and the title 20 Pisos de Historia (“20 Pisos ”) was chosen.

The session produced a number of suggestions for dramatis personae, many modeled on preexisting characters taken from other projects involving the team. For example, a focus of the sitcom was to be a gossipy security guard in the lobby named “Vázquez,” who was based on a character Logroño played in a comedy segment in a previous television special, Sunshine a la BBQ. “Soto,” an older woman living alone, was drawn from a character in another prior Logroño show, La Tripletta. These and like adaptations were used when the sitcom was aired.

After this discussion of ideas, Logroño asked Morales to write scripts for two of the first three episodes of 20 Pisos and Jiménez to write the third. Logroño’s recollection was that he framed the plots of all three, and Morales and Jiménez merely prepared dialogue to conform to Logroño’s storylines. By contrast, Morales claims that he wrote the two scripts at home, by himself, based on the general concepts aired at the meeting; similarly, Jiménez claims to have taken the general ideas from the session and fixed them in writing for the first time.

The writers delivered their scripts to Logroño, who retyped all three using special screen-writing software. On the cover pages, Logroño listed Morales as the author of episodes one and three and Jiménez as the author of episode two. Logroño claims to have made substantial changes in the scripts given to him, but the trail of drafts shows only minimal editing. And while Logroño appears to have co-written a number of later scripts for 20 Pisos, these reflected themes and characters set up in the first three episodes.

The first episode of 20 Pisos aired on the inaugural broadcast of the variety show, De Noche con Iris y Sunshine, on November 7, 1997. The show was produced by Creative Relief Corp. — a business wholly owned by Mojena — and broadcast in Puerto Rico by the WKAQ television station, an affiliate of the Telemundo network. The early episodes were successful and the variety show— and the 20 Pisos sitcom within it — became a recurring weekly program for almost two years (although renamed after Chacón left).

In December 1999, Logroño decamped from WKAQ for a rival television station— WAPA, an affiliate of Televicentro — where he starred in a new sitcom produced by Mass Productions (a company controlled by Logroño and his wife), titled El Condominio. The new show, which began airing on WAPA in March 2000, featured many of the same actors as 20 Pisos. The *468show retained the old characters and the sitcom unfolded in a virtually identical condominium setting, although new stories were developed and some new characters added.

On March 15, 2000, TMTV brought a copyright infringement action in federal district court against Logroño, his wife, and their production company, Mass Productions.1 The suit sought a declaratory judgment and various other legal and equitable remedies. Logroño filed an answer and counterclaim, asserting that he was the sole owner of the copyright to the outlines, scripts, and characters used for 20 Pisos and that he was owed royalty payments for the use of these properties by TMTV.

In due course, the district court granted summary judgment in the plaintiffs favor. TMTV, Corp. v. Mass Prods., Inc. (TMTV I), 345 F.Supp.2d 196 (D.P.R.2004). The court held that Morales and Jiménez had authored the first three scripts of 20 Pisos pursuant to valid work-for-hire agreements with TMTV’s predecessor in interest and that all later episodes of 20 Pisos were derivative of the first three. Id. at 204-08. As to infringement, the district court found 20 Pisos and El Condominio virtually identical. Id. at 213.

In December 2004, a month or so after this decision, Televicentro — the network on which El Condominio was broadcast— sought unsuccessfully to intervene and to have the partial summary judgment set aside. In August 2005, TMTV sought to consolidate its own suit with two others now pending before another district judge: an infringement suit by TMTV against Televicentro for broadcasting El Condominio and a suit against TMTV by a group of actors claiming copyright in the characters they had portrayed in 20 Pisos. This motion too was denied.

Thereafter, Logroño made his own failed attempt to undo the liability ruling against the defendants by claiming that the scriptwriters’ work-for-hire agreements with TMTV were invalid. TMTV, Corp. v. Mass Prods., Inc. (TMTV II), 453 F.Supp.2d 378, 382-83 (D.P.R.2006). He also sought, unsuccessfully, to forestall a trial on damages by arguing that a private intervening settlement of the suit by TMTV against Televicentro meant that any further award to TMTV would be an impermissible double recovery.

Prior to the trial, TMTV elected to claim only its actual damages (and not the defendants’ profits or statutory damages). See 17 U.S.C. § 504 (2006). TMTV’s expert witness, economist Michael Einhorn, estimated damages as at least $4.9 million, based on licensing fees TMTV allegedly would have earned but for the infringement by the defendants; this was measured in part by licensing fees actually earned by another comparable sitcom on Puerto Rican television. The jury awarded TMTV $772,079.29.

Following the verdict, the district court, over TMTVs objection, reduced the award by $700,000, representing the amount that TMTV had received in settlement from Televicentro. The court awarded TMTV prejudgment interest at the local-court rate of 5 percent per annum on the original jury verdict up through the date of the Televicentro settlement and, from then on, on the much smaller balance. Finally, the court denied TMTV’s request for attorneys’ fees as barred by the registration requirement of the Copyright Act, 17 U.S.C. § 412.

*469Both parties filed timely appeals, and we begin with Logroño’s challenges to the summary judgment — most importantly, his attack on the infringement ruling against him and his co-defendants. That ruling in turn required that TMTV show both ownership of a validly copyrighted work and improper copying of the protected elements of that work. Yankee Candle Co. v. Bridgewater Candle Co., 259 F.3d 25, 33 (1st Cir.2001). Our review of the summary judgment ruling is de novo.

Since TMTV’s claim to copyright runs through Jiménez and Morales, the first question is whether they or Logroño should be viewed as the creator of the original scripts. Jiménez and Morales say that once the condominium setting was agreed upon, they wrote the initial three scripts, framing the plots and dialogue and drawing in part on stock characters used in previous projects in which the team had participated. Logroño says that he outlined the plots and rewrote the scripts.

Ordinarily, a credibility contest means that the case must be tried but, in this instance, Logroño’s claim to have substantially rewritten the scripts is refuted by comparing the originals and the final product. As for Logroño’s supposed outlining of the plots, he has pointed to no writing — neither any written outline by him nor any notes taken by anyone at the original meeting — that reflects anything like an outline of the scripts. At best, Logroño’s version would mean that he suggested some general ideas for plots and possible stock characters for the initial scripts.

One of the axiomatic requirements for copyright is that the author’s work be “fixed.” 17 U.S.C. § 102(a). Copyright is a protection afforded to the definite expression of ideas but not the ideas themselves. See, e.g., Cmty. for Creative Non-Violence v. Reid, 490 U.S. 730, 737, 109 S.Ct. 2166, 104 L.Ed.2d 811 (1989); Erickson v. Trinity Theatre, Inc., 13 F.3d 1061, 1071 (7th Cir.1994). On this record, the district court properly found that there was no case on liability for a jury. Scott v. Harris, 550 U.S. 372, 380, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007).2

Logroño claims to have suggested the security guard and other characters; but nothing in the record indicates that they had the delineation needed to make them subject to copyright. Stock characters, like stock scenes afaire, are not subject to copyright protection. Nichols v. Universal Pictures Corp., 45 F.2d 119, 121 (2d Cir.1930), cert. denied, 282 U.S. 902, 51 S.Ct. 216, 75 L.Ed. 795 (1931); 1 P. Goldstein, Goldstein on Copyright § 2.7.2, at 2:91-93 (3d ed. 2005); cf. Coquico, Inc. v. Rodriguez-Miranda, 562 F.3d 62, 68 (1st Cir.2009). So even if Logroño suggested general plot ideas and stock characters, his claim to authorship of the scripts still fails.

Positing Jiménez and Morales as the authors, ownership must still be traced to TMTV. Although the copyright ordinarily vests in the initial author or authors, 17 U.S.C. § 201(a), TMTV claims that the three scripts fall within the work “made for hire” rubric, id. § 201(b), meaning that the original production company would be considered the author. The district court agreed and, the original production company having assigned its rights to TMTV, the court treated TMTV as the copyright own*470er. TMTV II, 453 F.Supp.2d at 381-82; TMTV I, 345 F.Supp.2d at 206-07.

Jiménez and Morales confirmed in depositions that this was their own oral understanding with the production company; but the statute requires express agreement in a signed written instrument, see 17 U.S.C. § 101 (definition); and the circuits are divided as to whether the language and policy require the writing before the creation or at least the completion of the work.3 But even assuming that the rights remained with the scriptwriters, before this case began they conveyed (in writing, 17 U.S.C. § 204(a)) whatever rights they had to TMTV. We may rely on this assignment even though the district court did not. Torres v. E.I. Dupont De Nemours & Co., 219 F.3d 13, 18 (1st Cir.2000).

Logroño next argues that he participated heavily in writing later scripts and that the district court’s conclusion that these scripts were derivative works of the first three was cursory and unsupported. See TMTV I, 345 F.Supp.2d at 207-08. But this was not his theory in the district court and so is forfeit, Santiago-Sepúlveda v. Esso Standard Oil Co. (P.R.), 643 F.3d 1, 1 (1st Cir.2011); nor, given the full scale development of the show in the first three scripts, would this new take be persuasive.

The second element in the liability claim is infringement, which requires proof that actual copying occurred and that the copying was so extensive that the two works are “substantially similar.” Segrets, Inc. v. Gillman Knitwear Co., 207 F.3d 56, 60 (1st Cir.), cert. denied, 531 U.S. 827, 121 S.Ct. 76, 148 L.Ed.2d 39 (2000). However, Logroño obviously had access to the original scripts and based El Condominio upon them, bringing with him a number of the actors. The initial episode of El Condominio mentioned the characters’ move to a new building; and the El Condominio advertising campaign invited viewers to reconnect with their favorite characters at a new location.

So copying is not in doubt and the issue is substantial similarity, which requires comparing the expressive elements of the copyrighted work contributed by the author — the components alone subject to copyright protection — with those of the allegedly infringing work. Yankee Candle, 259 F.3d at 33. No infringement claim lies if the similarity between two works rests necessarily on non-copyrightable aspects of the original — for example, “the underlying ideas, or expressions that are not original with the plaintiff.” Matthews v. Freedman, 157 F.3d 25, 27 (1st Cir.1998); accord Johnson v. Gordon, 409 F.3d 12, 19 (1st Cir.2005).

Infringement can occur where— without copying a single line — the later author borrows wholesale the entire backdrop, characters, inter-relationships, genre, and plot design of an earlier work.4 *471Imagine, for example, that the first Sherlock Holmes stories had been penned and copyrighted last year by Sir Arthur Conan Doyle, and Logroño had then written his own sequels carrying everything forward into a new plot. See, e.g., Anderson v. Stallone, No. 87-0592 WDKGX, 1989 WL 206431, at *8 (C.D.Cal. Apr. 25, 1989). Here, we agree with the district court that

the similarities between the two programs are so striking that there is no doubt that “El Condominio” is ... an unauthorized derivative work. In effect ... both programs are the same with only a difference in name and transmitted via a different channel. Specifically, the setting, character names, costumes, character interactions, comedy line, mood, [and] camera angles are almost identical in both sitcoms.

TMTV I, 345 F.Supp.2d at 213.

Logroño’s claimed differences between the two shows are not significant. For example, although the action in El Condominio unfolds largely in a condominium building lobby virtually identical to the lobby setting in 20 Pisos, the second sitcom also added some scenes in a laundry room and the handyman’s space; so, too, some characters are given more dialogue or minor variations in character traits {e.g., a new hearing problem). Such trivial modifications are not a defense. Cf. Coquico, 562 F.3d at 68; Matthews, 157 F.3d at 28.

Liability for copyright violation being established, we turn to damages. The jury awarded $772,079, but the district judge reduced this figure by $700,000 for the settlement payment TMTV received from Televicentro. Logroño argues that all of the damages awarded by the jury should have been erased by the settlement; TMTV, as cross-appellant, claims that none of the damages awarded by the jury should have been reduced by the settlement payment.

The acts of infringement by the defendants and by Televicentro are not identical but they are entangled: Logroño and his production company produced the infringing El Condominio programs and Televicentro broadcast them. In this case, TMTV’s theory of damages at trial was that the infringement diverted from TMTV potential licensing royalties; this was patently a claim for plaintiffs damages rather than defendants’ profits.5

We start with the defendants’ claims. The defendants make two different arguments for their view that the prior TMTV-Televicentro settlement barred any recovery at trial. The first is that the settlement constituted a release of claims against any available party who could be charged with infringement based on the same episode; the second, that the judgment entered upon the settlement is res judicata, precluding any further claims against the defendants.

Older common law doctrine tended toward the view that a settlement with one tortfeasor extinguished claims against other joint tortfeasors responsible for the same harm — although who was a joint tortfeasor and what constituted the same *472harm could be fraught questions; but the modern view is strongly in favor of treating a settlement with one joint tortfeasor as releasing the others only if the parties to the settlement so intended; otherwise, the settlement becomes an offset against later judgments if needed to avoid double recovery.6

The TMTV-Televicentro settlement, as an agreement, is arguably interpreted under Puerto Rico law, Great Clips, Inc. v. Hair Cuttery of Greater Bos., L.L.C., 591 F.3d 32, 35 (1st Cir.2010), although federal law might be thought to bear on whether a federal judgment incorporating a settlement should reduce a later federal judgment on a federal claim, Singer v. Olympia Brewing Co., 878 F.2d 596, 599-600 (2d Cir.1989), cert. denied, 493 U.S. 1024, 110 S.Ct. 729, 107 L.Ed.2d 748 (1990). But the question of which law applies is academic here, because both Puerto Rico and federal law follow the modern approach in looking to the intent of those who made the settlement.7

The defendants claim that TMTV’s acceptance of the settlement payment bespeaks an intent to relinquish claims against non-settling tortfeasors, absent an express reservation of rights. In some situations, a settlement with a joint tortfeasor might suggest such an intent, see Keeton, supra, § 49, at 335 — for example, a release discharging “any and all persons” potentially liable on the stated claims. But the opposite is often more plausible: frequently, the plaintiff gives a break to the first settling defendant, puts some money in the bank, and aims for a higher judgment against non-settling defendants. If the release were for all, a larger settlement would be demanded.

Here, the settlement itself was apparently oral and TMTV denies any such intent to release others besides Televicentro. The defendants — on whom the burden of proof falls to prove release, Fed. R.Civ.P. 8(c) — point to nothing to suggest that Televicentro had such an intent. Indeed, the defendants do not even suggest a plausible reason why Televicentro should have wished to pay a larger settlement in order to protect the defendants in this case.

Turning then to the judgment entered pursuant to the Televicentro settlement, it says that the $700,000 payment is made “in satisfaction of all causes of action identified in the Complaint [against Televicentro] and its amendments.” The claims in that complaint identified acts by Televicentro, including the broadcast of episodes of El Condominio, as alleged violations of TMTV’s copyright based on the original scripts; but the complaint asserted no claims against the present defendants based on their production of the shows.

A judgment has the meaning (objectively determined) intended by the *473court that framed and entered it, cf. Concilio de Salud Integral de Loíza, Inc. v. Perez-Perdomo, 625 F.3d 15, 19-20 (1st Cir.2010); but where the judgment is based on a settlement, courts may look to the intent of the settling parties, 18A C. Wright, A. Miller & E. Cooper, Federal Practice and Procedure § 4443, at 262 (2d ed. 2002); cf. Hermes Automation Tech., Inc. v. Hyundai Elecs. Indus. Co., 915 F.2d 739, 751 n. 10 (1st Cir.1990). Here, nothing shows that either the judge who entered the prior judgment or the parties to that case intended to release anyone else.

The result is the same if the settlement aspect is ignored and formal res judicata rules are applied to the judgment. Because no issues were adjudicated in the settled case, issue preclusion is irrelevant and the concern is only with claim preclusion. Classically, claim preclusion applies only where there is (1) a prior final judgment on the merits (2) on the same or sufficiently identical claims (3) between the same parties or their privies. Airframe Sys., Inc. v. Raytheon Co., 601 F.3d 9, 14 (1st Cir.2010).

Whether or not the defendants are treated as joint tortfeasors, they were not parties to the settled case and are not in privity with Televicentro. See Taylor v. Sturgell, 553 U.S. 880, 894, 128 S.Ct. 2161, 171 L.Ed.2d 155 (2008). Claim preclusion extends beyond parties and their privies only in unusual circumstances;8 consider, for example, a plaintiff, unsuccessful against an initial defendant, seeking to litigate identical claims against new but closely related defendants. Negrón-Fuentes v. UPS Supply Chain Solutions, 532 F.3d 1, 10 (1st Cir.2008).

Eliminating entirely the party or privy condition would require that whenever a plaintiff sued a defendant, the plaintiff also sue in the same action all other defendants against whom the plaintiff might have related claims. No such compulsory joinder requirement appears in the Federal Rules of Civil Procedure or elsewhere. Cf. Restatement (Second) of Judgments, supra, § 49. If such a rule were ever adopted— and this itself would be a debatable choice — fairness would require advance warning to potential plaintiffs and probably major qualifications on the requirement itself.

As for TMTV’s objection to the offset, the modern view, as already stated, is that an offset is conventional where needed to prevent recovering twice for the same harm. There are other solutions apart from a dollar-for-dollar reduction of the judgment by the amount of the settlement, including a more complex proportional regime that the Supreme Court has employed in admiralty cases. See McDermott, 511 U.S. at 211-17, 114 S.Ct. 1461; cf. Río Mar Assocs., 522 F.3d at 166-67 (Puerto Rico law). But neither side argues here that some other method should be used.

Rather, TMTV’s claim is that the reduction is not warranted because (it asserts) the settlement was for a different harm, in that Logroño prepared a derivative work, 17 U.S.C. § 106(2), while Televicentro publicly transmitted it, id. § 106(4). True, some of the infringing acts differed: Logroño acted and wrote scripts; the television company broadcast the programs. But the damages were (on TMTV’s own theory) the same harm: the lost royalties that could have been earned by TMTV for the production and broadcast of the El *474 Condominio programs if TMTV’s right had been respected' instead of infringed. Production and broadcast were two necessary, sequential steps needed to cause the same damage.

TMTV also says that the Televicentro settlement was for Televieentro’s profits rather than TMTV’s damages. Although damage liability is joint and several, Screen Gems, 453 F.2d at 554, liability for profits is several only, MCA, Inc. v. Wilson, 677 F.2d 180, 186 (2d Cir.1981).9 But nothing supports TMTV’s claim that the Televicentro settlement was, sub silentio, restitution of profits but not compensation of actual damages. Indeed, the damage expert in both cases was the same. In sum, the district court’s treatment of the settlement was correct.

The defendants argue lastly that the district court erred in its award of prejudgment interest (5 percent on the full jury damage award up to settlement and thereafter only on the balance). The defendants say that the Copyright Act does not authorize the award of prejudgment interest and that, if the statute does permit such an award, the equities do not favor an award here or at least the award should have been made at a lower rate of interest.

The Copyright Act does not expressly authorize prejudgment interest but authority for such an award can be inferred, where appropriate, from congressional purpose and general principles. Rodgers v. United States, 332 U.S. 371, 373, 68 S.Ct. 5, 92 L.Ed. 3 (1947). Here, as actual damages were calculated, TMTV was deprived of royalties that would have been due had the defendants produced the infringing programs under license. Prejudgment interest dating from the infringements compensated the plaintiff for the time value of monies it should have had — just as if a contract debt had not been paid on time.

Conversely, the defendants had the use of additional funds — whatever portion of their assets corresponded to such royalties — during the period before the judgment, so an award of prejudgment interest also avoids unjust enrichment. Cf. Frank Music Corp. v. Metro-Goldwyn-Mayer Inc., 886 F.2d 1545, 1552 (9th Cir.1989), cert. denied, 494 U.S. 1017, 110 S.Ct. 1321, 108 L.Ed.2d 496 (1990). Because the Televicentro settlement provided TMTV with most of the deficiency from that time forward, the defendants’ interest obligations on the reduced balance give them a slight windfall — although not at TMTV’s expense.

Our own case law treats an award of prejudgment interest in such eases as a matter for the informed discretion of the district court. See John G. Danielson, Inc. v. Winchester-Conant Props., Inc., 322 F.3d 26, 51 (1st Cir.2003).10 Given the range of circumstances and the mix of motives for such awards, this flexibility may make sense. Certainly on the pres*475ent facts there was no abuse of discretion in making the award or in limiting it in the fashion adopted by the district judge.

For post-judgment interest on federal judgments, a uniform federal rate applies by statute, 28 U.S.C. § 1961 (2006) (weekly average yield of one-year constant maturity treasury bills), but no default federal rate exists for prejudgment interest; here, the district judge applied the prevailing rate of 5 percent used in Puerto Rico courts for both pre- and post-judgment interest. P.R. Laws Ann. tit. 32, App. Ill, R. 44.3 (2000). The federal post-judgment section 1961 interest rate would have been approximately 0.5 percent when judgment was entered in January 2009.

The defendants claim that it was obligatory or at least more equitable to use the lower federal rate even though it is designed for post-judgment interest. The considerations are somewhat different where, absent a stay or bond, the successful plaintiff can effectively require prompt payment, see Fed.R.Civ.P. 64(b), and the present federal rate is depressed by current monetary policy designed to combat recession. The choice of the local rate, like the award itself, was within the sound discretion of the district judge. E.g., Colon Velez v. P.R. Marine Mgmt., Inc., 957 F.2d 933, 941 (1st Cir.1992). Five percent is by historical standards hardly exorbitant.

The last issue is attorneys’ fees. Ordinarily, the prevailing party in a copyright case can be awarded fees “[e]xcept as otherwise provided by” the statute. 17 U.S.C. § 505. Here, such an award was barred by statute because the copyright for the scripts was not registered in a timely fashion. Id. § 412; see, e.g., Knit-waves, Inc. v. Lollytogs Ltd., 71 F.3d 996, 1012 (2d Cir.1995) (vacating fee award). TMTV does not dispute that the statute barred such a recovexry on TMTV’s own claims against the defendants.

Instead, TMTV argues that it was also the prevailing party as to Logroño’s counterclaim, and so entitled to recover its attorneys’ fees attributable to defending against this claim. The argument — unpromising in any event — is forfeit, Dillon v. Select Portfolio Servicing, 630 F.3d 75, 80 (1st Cir.2011), having first been made in a motion for reconsideration of the district court’s denial of TMTV’s request for attorneys’ fees.

The district court’s judgment is affirmed. Each side is to bear its own costs on these cross-appeals.

It is so ordered.

OPINION OF THE COURT

I. Facts. .208

II. Procedural History . .209

III. The Direct Confusion Claim. O rH 03

A. The Lapp Test and the District Court’s Opinion H rH 03

B. The Test for Directly Competing Goods. (M rH 03

C. The District Court’s Methodology. lO rH 03

IV. Review of the District Court’s Analysis.216

A. Similarity of the Marks.216

1. Sight, Sound, Meaning.217

2. The Housemarks and the Disclaimer.218

3. The PTO’s Rejection of Victoria’s Secret’s Application.220

4. Summary.221

B. Strength of the Marks.221

1. Distinctiveness or Conceptual Strength.221

2. Commercial Strength of the Mark .224

3. Summary. 224

C. Product Similarity.224

D. Marketing and Advertising Channels .225

E. Sophistication of Consumers.225

F. The Intent of the Defendant.225

G. Actual Confusion.226

H. Combining the Factors.227

V. The Reverse Confusion Claim.227

A. Introduction.227

B. The Test for Reverse Confusion.229

1. The Factors that are the Same.229

2. Similarity of the Marks.229

3. Strength of the Marks.230

a. Commercial Strength.230

b. Distinctiveness or Conceptual Strength.231

*2064. The Intent of the Defendant .232

5. Factors Relating to Actual Confusion.233

6. Other Relevant Facts .234

7. Summary of the Test for Reverse Confusion .234

C. The District Court’s Opinion.234

D. Guidance for Remand.236

1. Introduction .236

2. Similarity of the Marks.236

3. Strength of the Marks.236

4. Intent.237

E. Summary.237

VI. Conclusion.238

BECKER, Chief Judge.

The critical question in this trademark infringement case, before us for the second time, is whether a typical consumer is likely to confuse MIRACLESUIT swimwear with THE MIRACLE BRA swimwear.1 The former is a product of Plaintiff A & H Sportswear Company (“A & H”), which manufactures ten percent of all swimsuits made in the United States. The latter is a product of Defendant Victoria’s Secret, the lingerie leviathan that recently entered the swimwear market. A & H filed suit in the District Court for the Eastern District of Pennsylvania claiming that The Miracle Bra swimwear mark violates the Lanham Act because it is confusingly similar to the Miraclesuit swimwear mark, which A & H registered first. A & H contends that: (1) consumers are likely to wrongly associate The Miracle Bra with A & H (the direct confusion claim2); or, in the alternative, (2) consumers are likely to think that Miraclesuit is a product of Victoria’s Secret (the reverse confusion claim).

During an extensive bench trial, A & H argued that Victoria’s Secret should be enjoined from using The Miracle Bra mark for swimwear. Finding a “possibility of confusion,” the District Court granted relief to A & H. Following an appeal to this Court that clarified that likelihood of eon-fusion (instead of possibility of confusion) was the correct standard, the District Court concluded that A & H had failed to show by a preponderance of the evidence that Victoria’s Secret’s The Miracle Bra swimwear mark created a likelihood of either direct or reverse confusion with the Miraclesuit product.

In Interpace Corp. v. Lapp, Inc., 721 F.2d 460, 463 (3d Cir.1983), this Court established a ten-factor test (the “Lapp” test) to determine the likelihood of confusion for direct confusion claims between goods that do not directly compete in the same market, but we have never decided what factors should be considered in the case of directly competing goods. The District Court therefore fashioned its own multi-factored test that approximates, but does not completely match, the Lapp test. In employing its test, the District Court acknowledged that the most important factor was the similarity of the marks, and determined that their overall commercial impressions were not similar.

The District Court placed particular emphasis on the fact that the The Miracle Bra mark typically appears alongside Victoria’s Secret’s housemark (the mark of the manufacturer), and that Victoria’s Secret uses a disclaimer with its mark that explicitly states that The Miracle Bra swimwear is unrelated to Miraclesuit or A *207& H. The court also concluded that Victoria’s Secret’s intent, the paucity of credible incidents of actual confusion, and the sophistication of the consumers weighed in favor of Victoria’s Secret, while the competitive proximity of the products and the strength of the Miraclesuit mark weighed in favor of A & H. Taking all this into account, the District Court found no likelihood of direct confusion between the marks, rejected A & H’s direct confusion claim, and, on the assumption that the disclaimer would continue to be used in an effective manner, denied injunctive relief. Notwithstanding that likelihood of confusion is an issue of fact subject to deferential review, the District Court’s disposition of the direct confusion claim has given rise to a number of important issues in this appeal.

A & H first contends that multi-factored tests, like the one used by the District Court, are inapplicable to competing goods, and that with directly competing goods a court should examine only the similarity of the marks. We disagree. Though a court need not look beyond the marks when goods are directly competing and the marks virtually identical, we conclude that the factors we have developed in the noncompeting goods context are helpful tools and should be used to aid in the determination of the likelihood of confusion in other cases. Our jurisprudence does not preclude this result, and the District Court did not err in taking other considerations into account.

Alternatively, A & H submits that the District Court erred as a matter of law by not following precisely the ten-factor Lapp test that we have previously developed to determine likelihood of confusion. It argues that by not considering certain factors that would clearly have weighed in favor of A & H, the District Court distorted the likelihood of confusion standard. We reject this argument for two reasons. First, we have developed the ten-factor Lapp test only as a guide. Although all of the factors can be useful, the Lanham Act does not require that they be followed precisely so long as the relevant comparisons suggested by the test are made. More importantly, although the District Court arguably used a creative test, all of the Lapp factors are integrated, in one way or another, into the court’s analysis. Therefore, even if the Lapp test were mandatory, A & H could not show prejudice.

A & H also argues that, regardless of the test to be used, the District Court clearly erred in resting its finding that there was no likelihood of confusion on Victoria Secret’s disclaimer. A & H asserts that it constitutes legal error even to consider such a disclaimer, but we disagree, seeing no reason to forbid courts from considering this kind of distinguishing feature. In this case, the District Court properly weighed the disclaimer. We also reject A & H’s assertion that the District Court erred in failing to give weight to the refusal of a Patent and Trademark Office (“PTO”) examining attorney to register The Miracle Bra mark for swimwear because of its similarity to Miraclesuit. Where, as here, the PTO examining attorney did not consider all the relevant factors, and lacked significant information that was available to the District Court, his determination need not be given weight. In fact, one legal error that we do find weighs in Victoria’s Secret’s favor: The District Court erred by relying almost entirely on its classification of A & H’s mark as “suggestive to arbitrary” in its determination that Miraclesuit was a conceptually strong mark deserving of a high degree of protection. In sum, we conclude that neither the District Court’s fact-finding nor its balancing of factors warrants reversal, and hence we will affirm its judgment on the direct confusion claim.

As for the reverse confusion claim, A & H challenges the District Court’s treatment as inadequate, and inconsistent with the method laid out in our leading reverse confusion case, Fisons Horticulture, Inc. v. Vigoro Industries, Inc., 30 F.3d 466 (3d *208Cir.1994). Fisons adopted the doctrine of reverse confusion, and used the Lapp factors to assess the likelihood of such confusion, with a few minor modifications. The District Court interpreted our precedents to require a two-step inquiry, engaging the Lapp factors only after an initial assessment that the disparity in commercial strength reached a high threshold. Because the threshold degree of commercial disparity that the court believed was required was not met, the court did not even examine whether there existed a likelihood of confusion.

After reviewing Fisons and our precedent, we are persuaded that the District Court erred in fashioning a two-step inquiry, and in failing to consider the Lapp factors as they apply to reverse confusion claims. We will demonstrate that the application of some of the factors changes in the reverse confusion context, and we conclude that, on the record before us, the judgment must be vacated with respect to the reverse confusion claim and the case remanded for further proceedings.

I. Facts3

The Miraclesuit bathing suit is made by A & H, which manufactures ten percent of all swimwear in the United States. The Miraclesuit is distributed by Swim Shaper, a division of Mainstream Swimsuits. The Miraclesuit is advertised as having a slimming effect on the wearer without using uncomfortable girdle-like binds. Its material purportedly smooths out middle body bulges and works with a flattering design to confuse the eye such that the wearer is advised, in advertising and in tags that generally accompany the product, that she will “[l]ook ten pounds fighter in 10 seconds[:] The ten seconds it takes to slip it on.” Miraclesuits also include tags indicating that they are Swim Shaper products.

Miraclesuits, which sell for between $50 and $100, come in both single pieces and bikinis. Many are equipped with push-up bras, shaping or underwire bras, or simple, unshaped bras. They are typically sold to trade buyers for department store sales and national mail-order catalogues, and, on two occasions, they were featured in the Victoria’s Secret catalogue. A & H received a trademark registration for the mark Miraclesuit in the fall of 1992. The District Court found that A & H has spent over $1.2 million to advertise the Miracle-suit in magazines and trade papers, and has received the equivalent of $1.5 million of advertising in “free publicity,” i.e., publicity in trade magazines, consumer columns, and the general press. The advertising and publicity campaign has been a success, and Miraclesuits constitute approximately ten percent of all of A & H’s sales.

While A & H is busy selling skinny waists and midriffs, Victoria’s Secret, the nation’s premier lingerie seller, has focused on instant enlargements of the bust.4 In 1993, Victoria’s Secret released The Miracle Bra, a padded push-up bra. Victoria’s Secret filed an application to register its The Miracle Bra trademark, and unleashed an avalanche of advertising and publicity, ultimately spending over $13 million on The Miracle Bra products. The campaign succeeded, and sales of The Mir*209acle Bra products have topped $140 million since they were first introduced.

In 1994, Victoria’s Secret’s trademark application for The Miracle Bra mark on lingerie was approved. Later that year, Victoria’s Secret moved The Miracle Bra mark into swimwear, and The Miracle Bra swimsuit and The Miracle Bra bikini started appearing in Victoria’s Secret cata-logues. (The Miracle Bra lingerie and swimwear are sold only in Victoria’s Secret stores and catalogues.). The cost of The Miracle Bra swimwear varies, but it is typically in the neighborhood of $70, and The Miracle Bra and Victoria’s Secret tags are prominently featured on all swimwear. This product also succeeded: The total sales of The Miracle Bra swimsuits reached $28 million by summer 1997. The last critical fact regarding this swimwear is that, as a result of this litigation, Victoria’s Secret has committed itself to using the following disclaimer with all promotion, advertising, and sales of The Miracle Bra: “The Miracle Bra Swimwear Collection is exclusive to Victoria’s Secret and is not associated with Miraclesuit by Swimsha-per.”

In 1995, after this litigation began, Victoria’s Secret applied to the PTO for a trademark for The Miracle Bra for swimsuits, bathing suits, and bikinis. Because it had previously conducted a search for The Miracle Bra as applied to lingerie, which had led it to the conclusion that The Miracle Bra did not threaten to infringe on other trademarks, Victoria’s Secret had not conducted a separate trademark search of The Miracle Bra trademark as it applied to swimwear. However, the PTO examining attorney denied Victoria’s Secret’s application due to its similarity to Miraclesuit because he determined that: (1) “Miracle” was the dominant feature of each mark; and (2) the product lines overlap. The denial was not appealed.

II. Procedural History

In December 1994, A & H filed a complaint against Victoria’s Secret, alleging direct and reverse trademark infringement, unfair competition, dilution, and unjust enrichment. The District Court held a bench trial on liability in the fall of 1995, and a damages and relief trial in 1996. The court found that Victoria’s Secret’s use of The Miracle Bra on its lingerie created no likelihood of confusion with A & H’s Miraclesuit. See A & H Sportswear Co. v. Victoria’s Secret Stores, Inc., 926 F.Supp. 1233, 1264 (E.D.Pa.1996) (“A & H I”). However, with regard to- Victoria’s Secret’s use of The Miracle Bra mark on swimwear, the District Court found a “possibility of confusion” with A & H’s Miracle-suit swimwear. See id. at 1269. Based upon this finding, it ordered Victoria’s Secret not to use The Miracle Bra tag with swimwear unless it was accompanied by the disclaimer: “The Miracle Bra (TM) swimwear collection is exclusive to Victoria’s Secret and not associated with Mira-clesuit (R) by Swim Shaper (R).” A & H Sportswear Co. v. Victoria’s Secret Stores, Inc., 967 F.Supp. 1457, 1482 (E.D.Pa.1997) (“A & H II"). The District Court also ordered that royalties on past and future sales of The Miracle Bra swimsuits be paid to A & H. See id. at 1483. Both parties appealed. Although the District Court granted a stay of its order pending appeal, Victoria’s Secret has continued to use the disclaimer voluntarily and has represented to us at oral argument that it will continue to do so regardless of the outcome of this litigation.

During the prior appeal, we heard the case en banc to consider the viability of the “possibility of confusion” standard originally used by the District Court. We rejected that standard, and remanded the case to the District Court to apply the traditional “likelihood of confusion” standard to the swimsuits’ marks. See A & H Sportswear Inc. v. Victoria’s Secret Stores, Inc., 166 F.3d 197, 206 (3d Cir.1999) (“A & H III”). The bulk of the en banc opinion is devoted to explaining the rejection of the “possibility of confusion” standard. However, we also commented on what has be*210come a central question in the current appeal, namely, whether the likelihood of confusion analysis differs for competing, as opposed to noncompeting, goods:

[T]he standard to be applied when goods that are the subject of a trademark infringement claim are directly competing, as is the situation with [Miraclesuit and The Miracle Bra swimwear], is different than that applied when the goods are not competing. The ten-factor test for likelihood of confusion between marks that are not competing, derived from Scott Paper Co. [v. Scott’s Liquid Gold, Inc.], 589 F.2d [1225,] 1229 [(3d Cir. 1978) ], is not required when the goods directly compete. In fact, we have said that “where the trademark owner and the alleged infringer deal in competing goods or services, the court need rarely look beyond the mark itself’.

Id. at 202 (quoting Interpace Corp. v. Lapp, Inc., 721 F.2d 460, 462 (3d Cir. 1983)). We also suggested that the District Court should, on remand, consider whether this case implicated the reverse confusion doctrine, noting the possibility that the later-entering mark of the junior user-in this case the larger, more powerful company (Victoria’s Secret)-might be overwhelming the mark of the prior, or senior, user (A & H), the phenomenon typifying reverse confusion.

Following our mandate, the District Court applied the “likelihood of confusion” test and found that there was no likelihood of direct confusion and no grounds for consideration of reverse confusion. It therefore denied the request for injunctive relief. See A & H Sportswear Co. v. Victoria’s Secret Stores, Inc., 57 F.Supp.2d 155, 178-79 (E.D.Pa.1999) (“A & H IV”). A & H appeals. We review the District Court’s factual determinations for clear error, but we give plenary review to its legal conclusions. See A & H III, 166 F.3d at 202-03. Likelihood of confusion is a factual question, see id., but legal principles govern what evidence may, or must, be considered by the District Court in reaching that conclusion, and also what standards apply to its determination. We review the denial of the request for injunc-tive relief for abuse of discretion. See Warner-Lambert Co. v. Breathasure, Inc., 204 F.3d 87, 89 n. 1 (3d Cir.2000).

III. The Direct Confusion Claim

We measure federal trademark infringement, 15 U.S.C. § 1114, and federal unfair competition, 15 U.S.C. § 1125(a)(1)(A), by identical standards. See A&H III, 166 F.3d at 202.5 To prove either form of Lanham Act violation, a plaintiff must démonstrate that (1) it has a valid and legally protectable mark; (2) it owns the mark; and (3) the defendant’s use of the mark to identify goods or services causes a likelihood of confusion. See Commerce Nat’l Ins. Servs., Inc. v. Commerce Ins. Agency, Inc., 214 F.3d 432, 437 (3d Cir.2000). The plaintiff bears the bur*211den of proof. See American Home Prods. Corp. v. Barr Labs., Inc., 834 F.2d 368, 371 (3d Cir.1987). It is undisputed that A & H owns Miraclesuit, and that it is a valid and legally protectable mark. Therefore, the questions in this case involve the delineation and application of standards for the evaluation of likelihood of confusion.

A. The Lapp Test and the District Court’s Opinion

A likelihood of confusion exists when “consumers viewing the mark would probably assume that the product or service it represents is associated with the source of a different product or service identified by a similar mark.” Dranoff-Perlstein Assocs. v. Sklar, 967 F.2d 852, 862 (3d Cir.1992) (quotation marks omitted). In Inteipace Corp. v. Lapp, Inc., 721 F.2d 460 (3d Cir.1983), we stated that when the goods involved in a trademark infringement action directly compete with each other, a court “need rarely look beyond the mark itself’ to determine the likelihood of confusion. Id. at 462. For noncompeting goods, we developed a non-exhaustive list of factors to consider in determining whether there is a likelihood of confusion between marks:

(1) the degree of similarity between the owner’s mark and the alleged infringing mark;

(2) the strength of the owner’s mark;

(3) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;

(4) the length of time the defendant has used the mark without evidence of actual confusion arising;

(5) the intent of the defendant in adopting the mark;

(6) the evidence of actual confusion;

(7) whether the goods, though not competing, are marketed through the same channels of trade and advertised through the same media;

(8) the extent to which the targets of the parties’ sales efforts are the same;

(9) the relationship of the goods in the minds of consumers because of the similarity of function;

(10) other facts suggesting that the consuming public might expect the prior owner to manufacture a product in the defendant’s market, or that he is likely to expand into that market.

Id. at 463 (citing Scott Paper Co. v. Scott’s Liquid Gold, Inc., 589 F.2d 1225, 1229 (3d Cir.1978)). Throughout the opinion we will refer to these factors as the “Lapp” factors.

The District Court recognized that we have not “explicitly elucidated what kind of factors should be considered in the case of directly competing goods.” A & HIV, 57 F.Supp.2d at 163. It chose not to use the Lapp test, but instead developed its own test, drawing on many of the same factors:

(1) strength of the plaintiffs mark;

(2) similarity between the marks;

(3) similarity of the products and the degree to which they directly compete with each other;

(4) marketing or advertising channels used;

(5) sophistication of consumers;

(6) defendant’s intent in selecting the mark; and

(7) incidents of actual confusion.

Id. A & H contends that the District Court erred in three ways in applying these factors. First, it argues that when goods are directly competing, a district court need only examine the similarity of the marks. Second, it submits that inasmuch as the District Court fashioned its own test, the Lapp factors it omitted would have tipped the balance in favor of finding a likelihood of confusion. Third, it contends that on any test the similarity of the marks creates a likelihood of confusion.

*212B. The Test for Directly Competing Goods

We have not previously settled upon a method that district courts should use to examine whether there exists a likelihood of confusion between directly competing goods. Uncertainty on this point in the district courts within the circuit, as demonstrated in the margin, counsels that we clarify this area of law.6

The multi-factored test rescribed above was developed to determine likelihood of confusion for noncompeting goods. See Versa Prods. Co., Inc. v. Bifold Co. (Mfg.) Ltd., 50 F.3d 189, 202 (3d Cir.1995). This genesis is apparent in the substance of the test: factor (7) directs a court to look at the markets of the goods “though not competing”; factor (9) directs that the court examine the “similarity of function”; and factor (10) directs that the court look at “other facts suggesting that the consuming public might expect the prior owner to manufacture a product in the defendant’s market, or that he is likely to expand into that market.” These factors are not apposite for directly competing goods: By definition, the goods are competing, their function is the same, and the senior and junior user are already in each other’s markets.

Nonetheless, despite the test’s etiology, the raison d’etre of the Lapp factors and the purpose of their development was to aid in the navigation of the difficult course of determining a “likelihood of confusion.” A district court should not be foreclosed from using any factors that it deems helpful in analyzing whether a likelihood of confusion exists between given products, whether or not they directly compete. Lapp’s suggestion that, for competing goods, the court “need rarely look beyond the mark itself,” Interpace Corp. v. Lapp, Inc., 721 F.2d 460, 462 (3d Cir.1983), may be an exercise in unjustified optimism; cases are often not easy, and courts will frequently need help in sorting out the likelihood of confusion. District courts within this circuit, as well as other appellate courts, have found that consideration of the Lapp factors (or their analogs from other circuits) can be quite useful for determining likelihood of confusion even when the goods compete directly. See, e.g., Barre-National, Inc. v. Barr Labs., Inc., 773 F.Supp. 735, 742 (D.N.J.1991); Banff, Ltd. v. Federated Dep’t Stores, Inc., 841 F.2d 486, 490 (2d Cir.1988).

Further, we recognize that many of the Lapp factors that are specifically applicable only to noncompeting goods are often used by courts to determine, in the first instance, whether goods are or are not directly competing. This makes the plaintiffs proposed prohibition on the examination of evidence beyond the marks themselves somewhat illusory. For instance, in AMF Inc. v. Sleekcraft Boats, 599 F.2d 341 (9th Cir.1979), the Court of Appeals for the Ninth Circuit concluded that two lines of recreational boats were not directly competitive because they were marketed to different consumers and were de*213signed for different uses. See id. at 348. This inquiry approximates Lapp factors (8) and (9), which look to the different functions of the products and the different target customers. Similarly, in Induct-O-Matic Corp. v. Inductotherm Corp., 747 F.2d 358 (6th Cir.1984), the court affirmed a district court’s conclusion that the parties were direct competitors based on an examination of the similarity of the parties’ customer base, geographical trading area, and products, see id. at 362, factors which approximate Lapp factors (7), (8), and (9). Indeed, in A & H I, the District Court first applied the Lapp test in analyzing A & H’s infringement claim regarding the use of The Miracle Bra mark on lingerie, and then observed that much of the same analysis could be used to determine whether The Miracle Bra swimsuit and the Miracle-suit shared the same market. See id. at 1266.7

The above observations lead us to conclude that the Lapp factors should be used both for competing and for noncompeting goods. As for those factors that specifically refer to noncompeting goods, we believe that, rather than force courts first to inquire as to whether or not the goods are directly competitive simply for the purpose of determining if those factors apply — and, presumably, applying Lapp factors to make this initial determination — factors (7), (9), and (10) of the Lapp test must be adapted to make them applicable whether the products directly compete or not.

In holding that the Lapp test is to be employed when examining both competing and noncompeting goods, we acknowledge that we have at times suggested that the method for comparing competing goods is different from that for noncompeting goods. In Williamsonr-Dickie Manufacturing Co. v. Davis Manufacturing Co., 251 F.2d 924 (3d Cir.1958), for example, we did not look beyond the marks themselves in upholding a trial court’s determination that the marks “Dickie Davis” and “Dick-ie’s” for boys’ clothing were confusingly similar. See id. at 926-27. Furthermore, in Lapp we stated that

[w]here the trademark owner and the alleged infringer deal in competing goods or services, the court need rarely look beyond the mark itself. In those cases the court will generally examine the registered mark, determine whether it is inherently distinctive or has acquired sufficient secondary meaning to make it distinctive, and compare it against the challenged mark.

721 F.2d at 462 (emphasis added). Additionally, in Fisons, we noted that the

showing of proof plaintiff must make for[likelihood of confusion] depends on whether the goods or services offered by the trademark owner and the alleged infringer are in direct competition.... Where the goods or services are not competing, the similarity of the marks is only one of a number of factors the court must examine to determine likelihood of confusion.

30 F.3d at 472-73. Finally, in A&H III, we stated that “the standard to be applied when goods ... are directly competing ... is different than that applied when the goods are not competing.” 166 F.3d at *214202. In each of these cases, we arguably implied that when the goods are competing, the similarity of the marks is the only relevant factor.

Nonetheless, we read our prior opinions on this subject to counsel only that a district court may, if it so chooses, examine only mark similarity for directly competing goods.

A & H urges us to take the quoted statement from A & H III to mean that a district court should not consider any factors at all beyond the similarity of the marks themselves. However, we believe that the word “standard” is different from the word “test.” The better reading of this statement is the literal one: The “standard” for deter mining when marks are likely to be confused is different when the marks directly compete, i.e., the factor regarding the similarity of marks may increase in importance, but it does not eliminate the other factors entirely. This proposition is not only unremarkable, it flows naturally from the Lapp test, which considers the competition and potential competition of the products in the marketplace. We have certainly never held that a court may not look beyond the marks themselves to determine likelihood of confusion. The quote from Lapp itself, res-cribed in our prior opinion in this case (“the court need rarely look beyond the mark itself’) presumes that a court might sometimes need to look beyond the mark. Lapp, 721 F.2d at 462.

We think that this inference is a sound one, for there appears to be no reason, statutory or otherwise, to impose the restrictive rule proposed by A & H. Indeed, the cases relied upon by A & H do no more than suggest that the Lapp inquiry may be unnecessary in some directly competing goods cases. See, e.g., A & H III, 166 F.3d at 202 (holding that in directly competing goods cases, the multi-factored test “is not required”). At all events, because we have never before been confronted with this question, any implication in our prior cases that courts are foreclosed from looking beyond mark similarity for competing goods would have been dictum.

Our conclusion that the Lapp factors may be used to determine the likelihood of confusion in cases of directly competing goods, at least when the marks are not identical, accords with the approaches of the Second, Sixth, Eighth, and Ninth Circuits, which also have approved using mul-ti-factored tests, developed for noncompetitive goods, in the competitive goods arena. See Banff, Ltd. v. Federated Dep’t Stores, Inc., 841 F.2d 486, 490 (2d Cir.1988); Jet, Inc. v. Sewage Aeration Sys., 165 F.3d 419, 421-22 (6th Cir.1999); Duluth News-Tribune v. Mesabi Publ’g Co., 84 F.3d 1093, 1096 (8th Cir.1996); Dr. Seuss Enters., L.P. v. Penguin Books U.S.A., Inc., 109 F.3d 1394, 1404 (9th Cir.1997).

As explained above, we do not hold that a District Court must use the factors. In fact our precedents suggest the opposite. If products are directly competing, and the marks are clearly very similar, a district judge should feel free to consider only the similarity of the marks themselves. See, e.g., Opticians Ass’n of Am. v. Independent Opticians of Am., 920 F.2d 187, 195 (3d Cir.1990) (“[v]ery little analysis” was necessary in a case where a splinter group of a larger organization continued to use the organization’s collective mark). Moreover, the court often need not apply each and every factor; when goods are directly competing, both precedent and common sense counsel that the similarity of the marks takes on great prominence.8 At all events, the factors are meant to be tools, not hurdles. On the other hand, when, as in this case, the degree of confusing similarity of the marks is not clear, see Barre National, Inc., 773 F.Supp. at 742, a court can turn to the other factors. Here, the District Court *215did not err in choosing to look beyond the marks themselves.

To summarize, we hold that whether or not the goods directly compete, the Lapp factors should be employed to test for likelihood of confusion. Therefore, likelihood of confusion for both competing and noneompeting goods should be tested with reference to the following:

(1) the degree of similarity between the owner’s mark and the alleged infringing mark;

(2) the strength of the owner’s mark;

(3) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;

(4) the length of time the defendant has used the mark without evidence of actual confusion arising;

(5) the intent of the defendant in adopting the mark;

(6) the evidence of actual confusion;

(7) whether the goods, competing or not competing, are marketed through the same channels of trade and advertised through the same media;

(8) the extent to which the targets of the parties’ sales efforts are the same;

(9) the relationship of the goods in the minds of consumers, whether because of the near-identity of the products, the similarity of function, or other factors;

(10) other facts suggesting that the consuming public might expect the prior owner to manufacture both products, or expect the prior owner to manufacture a product in the defendant’s market, or expect that the prior owner is likely to expand into the defendant’s market.

As we stress further below, the Lapp test is a qualitative inquiry. Not all factors will be relevant in all cases; further, the different factors may properly be accorded different weights depending on the particular factual setting. A district court should utilize the factors that seem appropriate to a given situation.

C. The District Court’s Methodology

A & H argues in the alternative that inasmuch as the District Court used a multi-factored test, it should have considered all of the Lapp factors, and that the “missing” factors would have favored A & H. It therefore contends that the court erred as a matter of law in fashioning its own test. However, the test used by the District Court is functionally the same as the Lapp test. A comparison of these factor lists reveals that the ostensibly missing Lapp factors appear to be incorporated into the District Court’s test. Lapp factor (9), “the relationship of the goods in the minds of consumers, whether because of the near-identity of the products, the similarity of function, or other factors” and Lapp factor (8), “the extent to which the targets of the parties’ sales efforts are the same,” are subsumed in the court’s factor (3), which considers the “similarity of the products and the degree to which they directly compete with each other.” Lapp factor (4), “the length of time the defendant has used the mark without evidence of actual confusion” is subsumed along with factor (6), “evidence of actual confusion,” within the District Court’s factor (7), “incidents of actual confusion.” Lapp factor (10), “other facts suggesting that the consuming public might expect the prior owner to manufacture both products, or expect the prior owner to manufacture a product in the defendant’s market, or expect that the prior owner is likely to expand into the defendant’s market,” is unmentioned, but also unnecessary, because the similarity of the goods and their competitive relationship was accounted for in the court’s factor (3).

A & H essentially argues that, had the Lapp test been applied as written, a greater absolute number of factors would have been decided in its favor, thus leading to a different outcome; for instance, the court’s factor (3), rather than counting in its favor a single time, would have been divided into at least two factors, each of which would have weighed in its favor. However, we *216have repeatedly insisted that the Lapp factors are not to be mechanically tallied, but rather that they are tools to guide a qualitative decision. See Fisons, 30 F.3d at 476 n. 11 (“The weight given to each factor in the overall picture, as well as its weighing for plaintiff or defendant, must be done on an individual fact-specific basis”).

The District Court did not engage in a simplistic quantitative comparison, but accurately understood the role of the factors:

No single factor is dispositive, and a finding of a likelihood of confusion does not require a positive finding on a majority of these factors. Instead, they are simply a guide to help determine whether confusion would be likely between the use of two contested trademarks on competing products. In addition to the listed factors, a court is free to consider other relevant factors in determining whether a likelihood of confusion exists.

A & H IV, 57 F.Supp.2d at 164 (citations omitted). Although it might promote clarity always to use the same factors in the same order, the District Court covered all the ground covered by the Lapp factors and did so with care. Therefore, we conclude, the court did not err in its use of the factors.

IV. Review of the District Court’s Analysis

Because we hold that the District Court’s methodology was functionally similar to the Lapp test, we will review the court’s decision on its own terms; i.e., using the multi-factored test that the District Court created. In the future, however, for the sake of consistency, district courts should employ the formulation of the test for confusion as set forth in Lapp and in Part III.B, supra.

A. Similarity of the Marks

The single most important factor in determining likelihood of confusion is mark similarity. See Fisons, 30 F.3d at 476. The test for such similarity is “ ‘whether the labels create the same overall impression when viewed separately.’ ” Id. (quoting Banff, Ltd. v. Federated Dep’t Stores, Inc., 841 F.2d 486, 492 (2d Cir. 1988)). Marks “are confusingly similar if ordinary consumers would likely conclude that [the two products] share a common source, affiliation, connection or sponsorship.” Id. Side-by-side comparison of the two marks is not the proper method for analysis when the products are not usually sold in such a fashion. Instead, an effort must be made to move into the mind of the roving consumer. See Ciba-Geigy Corp. v. Bolar Pharm. Co., Inc., 747 F.2d 844, 851 (3d Cir.1984) (affirming a district court’s conclusion that, in the trade dress context, “[r]ealistically, the likelihood of confusion cannot be assessed by a side-by-side comparison of the plaintiffs and defendant’s products. It is the overall physical appearance of defendant’s trade dress which is critical”); 3 J. McCarthy, Trademarks and Unfair Competition § 23:59, at 23-162 (4th ed. 2000) (“The law does not require that the reasonably prudent purchaser keep a handy file of photographs and labels which he or she must pull out to compare with the label of every product purchased.”).

The District Court acknowledged this principle, recognizing that “side-by-side comparison of the conflicting marks is improper if that is not the way buyers see the products in the market.” A & H IV, 57 F.Supp.2d at 167-68. However, the general rule that marks should be viewed in their entirety does not undermine the common-sense precept that the more forceful and distinctive aspects of a mark should be given more weight, and the other aspects less weight. See Country Floors, Inc. v. Partnership of Gepner and Ford, 930 F.2d 1056, 1065 (3d Cir.1991) (“When the dominant portions of the two marks ar e the same, confusion is likely.”). Applying these principles, the court looked at the sight, sound, and meaning of the marks, and considered the effect of the *217housemarks and Victoria’s Secret’s disclaimer.

1. Sight, Sound, Meaning

The District Court closely examined the sight, sound, and meaning of the competing marks and concluded that, purely on the aesthetic level, the marks alone are “somewhat distinct.” A&H TV, 57 F.Supp.2d at 167. Considering the sound of the marks, it noted that although they share the term MIRACLE, there are different numbers of syllables, and the last syllable of each is different. Moreover, Miraclesuit bleeds two words together while The Miracle Bra consists of three discrete words. In short, the District Court concluded that the two marks sound different. We agree. The court also properly noted that in previous cases, where courts had found a likelihood of confusion based on sound, a closer phonetic similarity was often present. See id. (citing Bell Publ’g Corp. v. Bantam Doubleday Dell Publ’g Group, 17 U.S.P.Q.2d 1634, 1637 (E.D.Pa.1990) (“DeZZ” and “Bell”)', Pocono Rubber Cloth Co. v. J.A. Livingston, Inc., 79 F.2d 446, 448 (3d Cir. 1935) (“swavelle” and “swavel”)).

The District Court recognized that other, less similar-sounding marks had been found to be confusingly similar, as in Williamson-Dickie Manufacturing Co. v. Davis Manufacturing Co., 251 F.2d 924, 926-27 (3d Cir.1958), where the court upheld the trial court’s determination that the marks “Dickie Davis” and “Dickie’s” were confusingly similar. However, it concluded that these cases could be distinguished because the presentations of the Miraclesuit and The Miracle Bra marks were sufficiently different to create an overall impression that was not confusingly similar, especially in light of Victoria’s Secret’s disclaimer and the use of the house-marks, as discussed infra Section IV.A.2.

Turning to the visual difference, the court noted that Miraclesuit is often accompanied by the “Look ten pounds lighter in ten seconds!” slogan, in various incarnations. Moreover, the court observed that:

The Miracle Bra is presented in either all capital block letters or in small capital letters with the letters, T, M, B alone in upper capital letters, while the Mira-clesuit has sometimes been advertised with the initial letter M capitalized and the rest of the mark in lower-case letters in italicized script or as one word with both the M and S capitalized.

A & HIV, 57 F.Supp.2d at 167. The court also determined that the meaning of the two marks was slightly different, as the descriptive word “bra” focused attention on the brassiere portion of the Victoria’s Secret product, rather than on the entire suit. See A & HIV, 57 F.Supp.2d at 166— 67 (stating that the term “bra” distinguishes the Victoria’s Secret product because it “identifies the particular feature of the garment where the ‘miracle’ is manifested”). This conclusion accords with the District Court’s finding in A & H I, which was upheld on appeal. See A&H Sportswear Inc. v. Victoria’s Secret Stores, Inc., 166 F.3d 191, 195 (3d Cir. 1999) (“[D]e-scriptive terms (such as ‘bra’ and ‘suit’) must be considered in assessing infringement.”).

Taking all of these elements into account, the District Court concluded that, without consideration of the role of the housemark and the disclaimer, “the Mira-clesuit mark and The Miracle Bra mark are somewhat distinct in sight, sound and meaning.” A & H IV, 57 F.Supp.2d at 168. A&H challenges this analysis. It begins by asserting that, because the dominant portion of each mark is the word “Miracle,” and the other words in each mark are generic (bra, swimwear, suit, the), the marks must be deemed confusingly similar. It grounds this analysis on Country Floors, Inc. v. Partnership of Gepner and Ford, 930 F.2d 1056 (3d Cir. 1991), in which we noted that it was relevant for purposes of comparison whether, in registering two marks, the PTO required certain descriptive elements of the *218mark to be disclaimed. Comparing “Country Tiles” and “Country Floors,” we held that because “floors” and “tiles” were discounted as generic in the application to the PTO, “the use of the word ‘Country’ by the Partnership is a material, although not necessarily a controlling, consideration on the Corporation’s claim that ‘Country Tiles’ logo directly infringed on the Corporation’s federal trademark.” Id. at 1065.

According to A & H, the same logic dictates that these marks are likely to be confused. Not only is the dominant portion of each mark “MIRACLE,” but the PTO examining attorney discounted the word “bra” in Victoria’s Secret’s PTO application to use The Miracle Bra for swimwear. Furthermore, A & H notes that in Fisons, in which “Fairway” and “Fairway Green” were up against each other, we took the district court to task for focusing too heavily on the differences between the marks instead of analyzing the overall impression. There, we held that “a subsequent user may not avoid likely confusion by appropriating another’s entire mark and adding descriptive or non-descriptive matter to it.” Id. at 477 (quoting 2 J. McCarthy, Trademarks and Unfair Competition § 23:15[8], at 23-102 (3d ed.1992)).

We agree with A & H that the words “the” and “bra” and “suit” are not particularly potent, but none of our cases has suggested a per se rule about the impact of generic terms within a nongeneric trademark; ultimately, the weight to be given each word is a judgment call, best suited to the fact-finder. Although the District Court might well have been more emphatic in stating its ultimate conclusions on this score, in our view, the court did not err, much less clearly err, in finding that the sight and sound were distinct, despite the fact that the differences involve generic elements.9

2. The Housemarks and the Disclaimer

Having found the marks only “somewhat distinct,” the District Court further concluded that Victoria’s Secret’s use of its housemark and disclaimer renders the marks dissimilar.10 As the District Court recognized, affixing a well-known housemark like that of Victoria’s Secret can help diminish the likelihood of confusion. See A & H IV, 57 F.Supp.2d at 168. We have previously acknowledged the importance of housemarks in comparing the “overall impression” of two marks. A & H Sportswear Inc. v. Victoria’s Secret Stores, Inc., 166 F.3d 191, 195 (3d Cir. 1999). In one of our previous decisions, we noted the relevance of the fact that both The Miracle Bra and Miraciesuit were sold with housemarks. See id.

Although A & H contends that Victoria’s Secret’s housemark should not have been a part of the similarity of marks analysis, the consideration of the effect of such marks has wide support in the case law. “Use of a strong, well-known mark as a part of a composite name reduces the likelihood that the remainder of the composite name will create a commercial impression distinct from that mark.” Four Seasons Hotels Ltd. v. Koury Corp., 776 F.Supp. 240, 247 (E.D.N.C.1991); see also W.W.W. Pharm. Co., Inc. v. Gillette Co., 984 F.2d 567, 573 (2d Cir.1993) (“[W]hen a similar mark is used in conjunction with a company name, the likelihood of confusion *219may be lessened.”); Astra Pharm. Prods, Inc. v. Beckman Instruments, Inc., 718 F.2d 1201, 1205 (1st Cir.1983) (“[Ojther-wise similar marks are not likely to be confused where used in conjunction with the clearly displayed name and/or logo of the manufacturer.”); Henri’s Food Prods. Co., Inc. v. Kraft, Inc., 717 F.2d 352, 355-56 (7th Cir.1983) (“Yet even Kraft agrees that a prominent housemark may also tend to lessen confusion.”). But see Frehling Enters., Inc. v. International Select Group, Inc., 192 F.3d 1330, 1337-38 (11th Cir.1999) (adding brand name did not lessen likelihood of confusion for marks of otherwise “striking similarity”); A.J. Canfield Co. v. Vess Beverages, Inc., 612 F.Supp. 1081, 1091 (N.D.Ill.1985) (“Vess’s use of its own VESS housemark in conjunction with Chocolate Fudge is not a defense to Canfield’s infringement claim, for the use of another’s trademark constitutes infringement with or without the use of the infringer’s housemark.”), ajfd, 796 F.2d 903 (7th Cir.1986).

Furthermore, and most importantly, the District Court concluded that the disclaimer used by Victoria’s Secret “create[d] a distinction” between the two marks. Therefore, it held that there was no similarity “solely based on the presumption that Defendants will continue to use the disclaimer when marketing The Miracle Bra swimwear.” A & H IV, 57 F.Supp.2d at 169. The court grounded its finding on the fact that

[ajlthough the positioning varied, the disclaimer often appeared in dark, black type against a light background at the lower left corner of the actual page on which The Miracle Bra Swimwear Collection was featured in the catalog, just above the telephone number for placing an order. The President of VS Cata-logue stated that since there are only a few seconds in which to capture a consumer’s attention, placement near the telephone number is particularly appropriate.

Id. (citations omitted). The disclaimer, the court concluded, tips the balance and makes it unlikely that consumers will be confused by the marks themselves.11

A & H argues that disclaimers cannot be considered when determining likelihood of confusion, relying heavily on United States Jaycees v. Philadelphia Jaycees, 639 F.2d 134, 142 (3d Cir.1981), in which we held that an injunction requiring an infringing organization to add the word “Philadelphia” to its use of the word “Jaycees” was too narrow a remedy for a Lanham Act violation. However, Jaycees does not control for two reasons. First, the addition of a geographical term, unlike the use of a housemark, is unlikely to vitiate confusion, as one could easily conclude that the Philadelphia Jaycees was a subgroup of the larger national organization. In this case, the contest is not between Miraclesuit and Philadelphia Miraclesuit, a situation that would probably cause confusion, but rather is between two different names, one of which has a prominent disclaimer highlighting the difference. Second, our decision in Jaycees addressed the propriety of a remedy, and was premised on a finding of a Lanham Act violation, whereas in this case, we analyze the disclaimer and house-mark to determine whether there was any violation at all.

In sum, the District Court committed no legal err or in considering the disclaimer. We are also satisfied that the District Court committed no factual error in concluding that Victoria’s Secret’s disclaimer helped to dispel potential consumer confusion between the Miraclesuit and The Miracle Bra swimwear.

*2203. The PTO’s Rejection of Victoria’s Secret’s Application

Finally, A & H argues that the District Court erred in failing to give weight to the fact that the PTO attorney rejected Victoria’s Secret’s application to register The Miracle Bra for “swimsuits, bathing suits and bikinis” on the ground that there was a likelihood of confusion between its mark and Miraclesuit.12 A & H does not argue that a PTO determination is controlling, but rather that it is important evidence that should be given considerable weight.13

There is some jurisprudence, although not in Third Circuit case law, suggesting that a PTO determination that marks are likely to be confused should be given weight as a matter of law. See, e.g., Guardian Life Ins. Co. of Am. v. American Guardian Life Assur. Co., 943 F.Supp. 509, 523 (E.D.Pa.1996) (“While not disposi-tive of the issue of likelihood of confusion, the PTO’s refusal to register Defendant’s marks is entitled to substantial consideration by this Court.”); Driving Force, Inc. v. Manpower, Inc., 498 F.Supp. 21, 25 (E.D.Pa.1980) (same); Syntex Labs., Inc. v. Norwich Pharm. Co., 437 F.2d 566, 569 (2d Cir.1971) (refusal of Patent Office to register a mark is “entitled to great weight”).

On the other hand, the Ninth Circuit has held that a preliminary determination by a low-level PTO administrator should not be accorded much weight, especially where the PTO officer did not have access to the full panoply of information that might inform a likelihood-of-confusion analysis. See Carter-Wallace, Inc. v. Procter & Gamble Co., 434 F.2d 794, 802 (9th Cir. 1970). The court explained:

Any such determination made by the Patent Office under the circumstances just noted must be regarded as inconclusive since made at its lowest administrative level.... The determination by the Patent Office is rendered less persuasive still by the fact that the Patent Office did not have before it the great mass of evidence which the parties have since presented to both the District Court and this court in support of their claims.

Id. (citation omitted). Moreover, other courts have held that a court need not *221defer to the patent office when there is relevant evidence not considered by the office that informs the analysis. See, e.g., Marketing Displays, Inc. v. Traffix Devices, Inc., 200 F.3d 929, 934 (6th Cir. 1999).

We find Carter-Wallace and Marketing Displays persuasive, and conclude that, although an initial PTO determination by an examining attorney may be considered, it need not be given weight when the PTO attorney did not review all the evidence available to the District Court. In A & H I, 926 F.Supp. at 1255, the District Court concluded that even a preliminary PTO determination should be given “substantial weight,” but it gave no weight to the PTO’s decision in its most recent opinion. It was not clear error to refuse to do so. As in Carter-Wallace, the PTO in this case was making a low-level preliminary determination, and did not have the benefit of the complete record before the District Court. Furthermore, the PTO attorney’s decision was eonelusory, not searching or analytical. See note 9, supra. Although we prefer to avoid conflicts with the PTO, we do not think that such a decision needed to receive deference here, where the District Court’s conclusion that the marks were not confusingly similar relied upon the housemarks and Victoria’s Secret’s disclaimer, matters apparently not considered by the examining attorney.

4. Summary

In sum, we conclude that the District Court did not clearly err in concluding that the marks themselves were not confusingly similar, considering their overall commercial impression, and did not make an error of law in choosing to consider Victoria’s Secret’s housemark and the disclaimer, or in refusing to give weight to the decision of the PTO attorney.

B. Strength of the Marks

With respect to Lapp factor (2),“the strength of the owner’s mark,” the District Court applied the Fisons test, which measures mark strength by “(1) the distinctiveness or conceptual strength of the mark; and (2) the commercial strength or marketplace recognition of the mark.” A&H1V, 57 F.Supp.2d at 164. The first prong of this test looks to the inherent features of the mark; the second looks to factual evidence of “marketplace recognition.” See Fisons, 30 F.3d at 479. The District Court concluded that the mark Miraclesuit ranged “conceptually” from suggestive to arbitrary, and that evidence of A & H’s advertising expenditures demonstrated that Miraclesuit had commercial strength, as well. Level of distinctiveness and mark strength are factual determinations that we review for clear error, see Ford Motor Co. v. Summit Motor Prods., Inc., 930 F.2d 277, 292 n. 18 (3d Cir.1991), but our review of legal conclusions is plenary.

1. Distinctiveness or Conceptual Strength

In order to determine whether a mark is protectable as a trademark, marks are divided into four classifications: (1) generic (such as “DIET CHOCOLATE FUDGE SODA”); (2) descriptive (such as “SECURITY CENTER”); (3) suggestive (such as “COPPERTONE”); and (4) arbitrary or fanciful (such as “KODAK”). See Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 768, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992).14 Arbitrary or fanciful marks use terms that neither describe nor suggest anything about the product; they “bear no logical or suggestive relation to the actual characteristics of the goods.” A.J. Canfield, 808 F.2d at 296 (citation omitted). Suggestive marks require con*222sumer “imagination, thought, or perception” to determine what the product is. Id. at 297. Descriptive terms “forthwith convey[ ] an immediate idea of the ingredients, qualities or characteristics of the goods.” Id. Generic marks are those that “function as the common descriptive name of a product class.” Id. at 296. In order to qualify for Lanham Act protection, a mark must either be suggestive, arbitrary, or fanciful, or must be descriptive with a demonstration of secondary meaning. See id. at 297. Generic marks receive no protection; indeed, they are not “trademarks” at all. See id. at 305.

Under the Lanham Act, stronger marks receive greater protection. See, e.g., Versa Prods. Co., Inc. v. Bifold Co. (Mfg.) Ltd., 50 F.3d 189, 203 (3d Cir.1995) (observing that stronger marks carry greater recognition, and that therefore a similar mark is more likely to cause confusion). Although the conceptual strength of a mark is often associated with the particular category of “distinctiveness” into which a mark falls (i.e., arbitrary, suggestive, or descriptive), that is not the only measure of conceptual strength. This is because the classification system’s primary purpose is to determine whether the mark is protectable as a trademark in the first place — that is, to determine whether consumers are likely to perceive the mark as a signifier of origin, rather than as a mere identification of the type of product. See A.J. Canfield, 808 F.2d at 296; Banff, Ltd. v. Federated Dep’t Stores, Inc., 841 F.2d 486, 489 (2d Cir.1988). The classification of a mark as arbitrary, suggestive, or descriptive is only secondarily used to deter mine the degree of protection a mark should receive once protectability has been established. These two inquiries—whether a mark is, in fact, a trademark, versus how much protection the mark should receive—are often identical, but they do not have to be. See Plus Prods, v. Plus Discount Foods, Inc., 722 F.2d 999, 1005 (2d Cir.1983) (“Although this classification system is a helpful tool [for determining strength] ... it is not determinative.... ”).

Suggestive or arbitrary marks may, in fact, be “weak” marks, particularly if they are used in connection with a number of different products. For instance, in H. Lubovsky, Inc. v. Esprit de Corp., 627 F.Supp. 483 (S.D.N.Y.1986), the court concluded that the mark “Esprit,” though suggestive, was, in fact, a “weak” mark. See id. at 487. Similarly, common marks like “Arrow,” though certainly not particularly descriptive of the underlying product, have been held to be “weak” marks. See Arrow Distilleries, Inc. v. Globe Brewing Co., 117 F.2d 347, 351 (4th Cir.1941). “Self-laudatory” marks like “Sure,” see Procter & Gamble Co. v. Johnson & Johnson, Inc., 485 F.Supp. 1185, 1196 (S.D.N.Y.1980), “Super Duper,” see S.M. Flickinger Co., Inc. v. Beatrice Foods Co., 174 U.S.P.Q. 51, 56 (T.T.A.B.1972), “Plus,” see Plus Prods., 722 F.2d at 1005, or, in this case, “Miracle,” are generally held to be weak marks. See 2 McCarthy, supra, § 11:81, at 11-146 to 147.15 Thus, the classification of a mark as “descriptive” or “arbitrary” for the purpose of determining whether it receives trademark protection at all — though a useful guide in assessing the strength or weakness of a mark — is not dispositive. See Express Servs., Inc. v. Careers Express Staffing Servs., 176 F.3d 183, 186 (3d Cir.1999) (observing that classification of distinctiveness is “useful” in determining conceptual strength (citing Banff, Ltd., 841 F.2d at 491)).

The District Court originally determined that the MIRACLE part of Miraclesuit was fanciful because it does not describe or reveal anything about the product. However, the court believed that the SUIT part was generic because it merely describes the qualities of the product. In sum, taking the mark’s parts together, the court concluded that Miraclesuit “ranges *223from suggestive to arbitrary” because it requires some consumer imagination to determine its meaning, and therefore that the mark should receive the highest level of protection. A & H IV, 57 F.Supp.2d at 165.

For reasons we will presently explain, we conclude that the District Court’s categorization of the Miraclesuit mark was clearly erroneous, and that the term Mira-clesuit does not rise to the arbitrary level but is, at best, merely suggestive. Further, we believe that the Court erred as a matter of law in holding that its categorization of the mark was dispositive of the inquiry into the mark’s strength. Because both of these conclusions were relevant to the District Court’s determination that the mark was entitled to a high level of protection, we review each separately.

The District Court concluded that the Miraclesuit mark ranged from “suggestive to arbitrary” because it believed that the word MIRACLE did not reveal anything about the product. To the contrary, we believe that both parts of the mark describe or suggest something about the product: The word “suit” has many meanings, but “bathing suit” is a frequent and familiar one. The word MIRACLE, unlike, say, XEROX, indicates the effect that the product is supposed to have on the user or wearer, and is ultimately merely a declaration of praise. Just as a “miracleworker” is a person who works miracles, a “miraclesuit” is a suit that works miracles. Whether the slimming effect is literally miraculous may be subject to debate, but the mark itself communicates something about the product. Suggestive or descriptive marks are, of course, protected, but, depending on surrounding circumstances (discussed below), they may receive lesser protection than arbitrary marks.

We also believe that the District Court committed an error of law in its automatic conclusion that, because the Miraclesuit mark “range[d] from suggestive to arbitrary,” it therefore should receive the highest level of protection, thus dismissing efforts by Victoria’s Secret to show that the Miraclesuit mark is conceptually weaker because MIRACLE has been used by several other companies. MIRACLE is used in other apparel markets, such as hosiery, children’s wear, ready-to-wear, and maternity wear. The court held that these other uses were irrelevant because the term MIRACLE was not used specifically in conjunction with swimwear. See A & H TV, 57 F.Supp.2d at 165 n. 16. Although the wide use of a term within the market at issue is more probative than the wide use of a term in other markets, see Fisons, 30 F.3d at 479, the extensive use of the term in other markets may also have a weakening effect on the strength of the mark.

For example, in Sun Banks of Florida, Inc. v. Sun Federal Savings & Loan Association, 651 F.2d 311, 316-17 & n. 8 (5th Cir.1981), the court gave special weight to the fact that 25 competing financial institutions used the word “sun” in their titles, but also noted that over 4,400 Florida businesses used the term. The Sun Banks court thus clearly considered extensive use in other markets in its assessment of the weakness of the contested term. See also Triumph Hosiery Mills, Inc. v. Triumph Int’l Corp., 308 F.2d 196, 199 n. 2 (2d Cir.1962) (“The mark ‘Triumph’ is a so-called weak mark, i.e. it has been used many times to identify many types of products and services.”). The relevance of such other uses of similar marks is apparent; if a consumer is aware that a particular mark, like “Triumph” or “Ace,” is often used to designate a variety of products made by a variety of manufacturers, that consumer will be less likely to assume that in a particular case, two individual products, both with the mark “Triumph,” come from the same source. See Steve’s Ice Cream v. Steve’s Famous Hot Dogs, 3 U.S.P.Q.2d 1477, 1479 (T.T.A.B.1987) (“[T]he numerous third-party uses [of Steve’s] demonstrate that the purchasing public has become conditioned to recognize *224that'many businesses ... use the term ... and ... is able to distinguish between these businesses based on small distinctions among the marks.”); cf. S.C. Johnson & Son, Inc. v. Johnson, 116 F.2d 427, 430 (2d Cir.1940) (“When all is said, if a man allows the good will of his business to become identified with a surname so common as Johnson, it is fair to impose upon him some of the risk that another Johnson may wish to sell goods not very far afield.... ”).

A & H notes that in Fisons we stated that “[wjhile other registrations and uses of Fairway for related products and services would make the mark less strong if they were in the same market, their use in different markets and for products and services that are not closely related does not necessarily undermine Fisons’ claim of strength.” 30 F.3d at 479. We read this sentence to mean no more than it says, and not, as A & H urges, as a per se rule against considering the use of the mark in different markets and for unrelated products. The short of it is that, whatever category of distinctiveness into which the mark falls, the multiple uses of MIRACLE in other markets is relevant to a determination of A & H’s mark’s strength.

2. Commercial Strength of the Mark

Turning to the second part of the test, the District Court determined that the Miraclesuit mark has a high level of commercial strength. It based its finding on the fact that A & H had spent so much money on advertising, and that “their efforts undoubtedly resulted in increased public recognition.” A & H TV, 57 F.Supp.2d at 165. The court also noted that A & H’s swimwear sales have steadily increased. Taking these two indications of commercial strength into account, the District Court concluded that the commercial strength inquiry weighed in favor of A & H. Although evidence of money spent does not automatically translate into consumer recognition, it is clearly relevant, and hence the District Court’s methodology and conclusion does not appear to be in error or to involve a misapplication of the law.

3. Summary

The District Court erred in concluding that the Miraclesuit mark contained an arbitrary component, and in concluding that the mark Miraclesuit was conceptually strong solely on the basis of its categorization. However, if anything, the District Court’s error weighed in A & H’s favor by leading the court to reject evidence that the Miraclesuit mark was weaker than A & H claimed. Therefore, the District Court’s error does not affect our affirmance of the District Court’s conclusion that there was no likelihood of direct confusion between Miraclesuit and The Miracle Bra.

C. Product Similarity

In discussing product similarity, the District Court first noted that although the products were both swimsuits, they have slightly different functions: The Miraclesuit is designed and advertised to make the figure appear slim, whereas The Miracle Bra suit is designed and advertised to enhance cleavage. However, the court also noted that most Miraclesuits have bras — some even have cleavage enhancing bras — and that a few Miracle Bra suits have lower body control. A review of these facts led the District Court to conclude that the Miraclesuit and Miracle Bra suits were “somewhat interchangeable.” A & H TV, 57 F.Supp.2d at 170. Therefore, it concluded, the product similarity factor favored A & H.

Given the evidence that, although both products may have a primary bust or midriff focus, both products also have attributes that enhance both aspects of the female figure, we do not think that the District Court clearly erred in making this finding. Furthermore, there was evidence that both products pitch themselves as “improving” female shapeliness through swimwear, and although the means of improvement are different and *225the products have different foci, the two suits are similar. In sum, while the products are not identical, the court committed no error of law here, and its conclusion on product similarity is supported by the record.

D. Marketing and Advertising Channels

The District Court concluded that the channels of sale and advertising of the products overlap. It noted that both the price range for the suits and their means of sale are similar. It observed that the Miraelesuit has been advertised in in-store promotions, magazines, department store bill enclosures, and press kits to buyers, editors, and publishers, and that The Miracle Bra suits have been marketed in point-of-sale promotions in Victoria’s Secret stores, and through press kits and print and television advertising. The court held that the channels of trade were similar, in that Miraelesuit is sold to department stores and catalogues that compete with Victoria’s Secret Stores and Victoria’s Secret Catalogue. Although the court noted that The Miracle Bra suits are never sold outside of the Victoria’s Secret store or catalogue, and that “[s]uch fine distinctions between the channels in which the products are sold may weigh in favor of diminishing a likelihood of confusion,” A & H IV, 57 F.Supp.2d at at 171, it concluded that the “conditions of purchase” were similar, and weighed this factor in favor of A & H.

Although there are significant differences in trade channels — notably, that Victoria’s Secret sells its product only in its own stores and catalogues — perfect parallelism will rarely be found. Furthermore, the Miraelesuit has twice been sold in Victoria’s Secret catalogues. Thus, the District Court did not clearly err in its factual conclusion that channels of marketing and advertising weigh in favor of A & H.

E. Sophistication of Consumers

The District Court next considered the “sophistication of consumers,” which is the functional equivalent of Lapp factor (3), “the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase.” The court concluded that the products’ consumers were likely to be sophisticated, and weighed this factor in favor of the defendants. Three considerations drove this decision. The court relied on cases holding that buyers of women’s apparel are sophisticated purchasers, and on the fact that A & H had presented no evidence that its consumers were not sophisticated. The court also noted that the “entire success” of each brand “relies on the premise that consumers will discern the slimming effect and cleavage enhancement features of their respective swimsuits.” A & H TV, 57 F.Supp.2d at 172.

A & H submits that the District Court erred in reaching this conclusion. It points to Victoria’s Secret’s admission during this litigation that the attention of consumers must be captured within a few seconds. See A & H II, 967 F.Supp. at 1466. However, there are no hard and fast rules for this determination, and the celerity with which one makes a decision does not correlate exactly with the extent of one’s discrimination in matters of taste. The District Court’s conclusion that consumers will be discriminating in their selection of swimwear, whether one-piece or bikinis, rings especially true. We find no error in the District Court’s conclusion that this factor weighs in favor of Victoria’s Secret.

F. The Intent of the Defendant

Before discussing this factor, it is important to reiterate the purpose of the “intent” inquiry in the Lapp analysis. Intent is relevant to the extent that it bears on the likelihood of confusion. We have held that a defendant’s mere intent to copy, without more, is not sufficiently probative of the defendant’s success in causing confusion to weigh such a finding in the plain*226tiffs favor; rather, defendant’s intent will indicate a likelihood of confusion only if an intent to confuse consumers is demonstrated via purposeful manipulation of the junior mark to resemble the senior’s. See Versa Prods. Co., Inc. v. Bifold Co. (Mfg.) Ltd., 50 F.3d 189, 205-06 (3d Cir.1995).16 Courts’ willingness to infer the efficacy of an intent to confuse is thus somewhat punitive, due in part to a willingness to believe that a closer nexus exists between an intent to deceive and success in that effort than exists between an intent to copy and success in causing confusion. See id. at 205-07. Some degree of bad faith is also relevant to the selection of an appropriate remedy. See A & H III, 166 F.3d at 208-09.

The District Court found that Victoria’s Secret had not intended to confuse customers when it began using The Miracle Bra in conjunction with its swimwear line. The record supports the conclusion that Victoria’s Secret brought the mark into swimwear because of its success in lingerie. The court thus properly found that Victoria’s Secret had not expanded The Miracle Bra into swimwear in an effort to ride on A & H’s good name.

A & H argues that the court should have weighed the intent factor in its favor because when The Miracle Bra was registered for lingerie, Victoria’s Secret’s counsel learned that there were several preexisting MIRACLE marks, including A & H’s Miraclesuit. However, the court credited evidence that in 1994, when Victoria’s Secret attempted to register The Miracle Bra for “swimsuits, bathing suits and bikinis,” Victoria’s Secret Stores assumed that its original trademark search had uncovered no confusingly similar registrations. See A & H IV, 57 F.Supp.2d at 173. The court also credited the evidence that no one at Victoria’s Secret Stores responsible ior the expansion of The Miracle Bra to swimwear had ever heard of the Miracle-suit or knew that Victoria’s Secret Cata-logue had ever sold Miraclesuits. Further, the court found that Victoria’s Secret Cat-alogue had chosen to expand the mark into swimwear for “legitimate reasons.” Id. The PTO attorney’s denial of registration occurred after this suit had been filed. See id. The District Court held, therefore, that it was “entirely credible” that no one at Victoria’s Secret Stores knew about Mi-raclesuit and that “[a]t best, this lack of communication was a result of bureaucratic carelessness and was not intentionally done for the purpose of profiting from the notoriety of Plaintiffs’ Miraclesuit mark.” Id. at 173-74. We discern no clear error or misapplication of law in these findings, or in the District Court’s weighing of the intent factor in Victoria’s Secret’s favor.

G. Actual Confusion

The District Court found that, although there was some evidence of actual confusion, that evidence was insufficient to permit weighing this factor in A & H’s favor. A & H produced an article in Women’s Wear Daily mentioning “the introduction of the Miracle Swimsuit in the upcoming Victoria’s Secret Catalog.” A & H TV, 57 F.Supp.2d at 174. An A & H sales agent testified that a professional swimwear buyer asked him if A & H carried The Miracle Bra swimsuit; that a professional swimwear representative asked if the two were related; that a former buyer asked if the Miraclesuit had that push-up element she had hear d so much about; and that a buyer asked him for an appointment to see The Miracle Bra fine. See id. A & H also presented testimony that one of its own public relations agents thought that A & H made both Miraclesuit and The Miracle Bra; that a buyer stated that she had hear d of the *227Miraelesuit as a suit that enhanced the bust; and that an A & H receptionist had received two inquiries concerning The Miracle Bra. See id.

The District Court, while not explicitly discrediting this evidence, viewed it with great skepticism, given the interested sources and the inability to cross-examine the supposedly confused individuals.17 Furthermore, it concluded that, even if it credited all the submissions, the evidence of actual confusion was isolated and idiosyncratic. See id. at 175. This decision is supported by the case law. “Ownership of a trademark does not guarantee total absence of confusion in the marketplace. Selection of a mark with a common surname naturally entails a risk of some uncertainty and the law will not assure absolute protection.” Scott Paper Co. v. Scott’s Liquid Gold, Inc., 589 F.2d 1225, 1231 (3d Cir. 1978). Although some cases hold that, given the difficulty of proving actual confusion, relatively little showing on the part of the plaintiffs is required, see, e.g., World Carpets, Inc. v. Dick Littrell’s New World Carpets, 438 F.2d 482, 489 (5th Cir.1971), other cases warn against using isolated instances of confusion to buttress a claim, see, e.g., Amstar Corp. v. Domino’s Pizza, Inc., 615 F.2d 252, 263 (5th Cir.1980). It is within the District Court’s discretion to consider the facts, and weigh them. In our view, the District Court’s conclusion as to the absence of actual confusion was supported by the record, and hence we will not disturb its decision to weigh this factor in favor of Victoria’s Secret.

H. Combining the Factors

As the foregoing discussion demonstrates, the District Court carefully considered the relevant factors weighing for and against a finding of likelihood of confusion. Most importantly, the court determined that, although the overall commercial impressions of the marks are only “somewhat distinct” if one looks just to sight, sound, and meaning, the fact that Victoria’s Secret uses a housemark and a disclaimer tips the balance and makes this factor, the similarity of the marks themselves, weigh in favor of Victoria’s Secret. This factor was essential to the District Court’s ultimate finding of fact; it found that “solely” on the assumption that the disclaimer would continue to be used, the marks were not similar. As to the other factors, the court found that: (1) the products; and (2) the channels of marketing were sufficiently similar to weigh those factors in favor of A & H, but that: (3) the consumers were sophisticated; (4) Victoria’s Secret exhibited no culpable intent in selecting its mark; and (5) there was insufficient credible evidence of instances of actual confusion, all of which weighed in favor of Victoria’s Secret. We find no reason to disturb any of these findings. Finally, the court also concluded that the mark Miraelesuit merited a high level of protection, and weighed this factor in favor of A & H. Though we believe that the District Court could have been more emphatic in its finding that the marks were dissimilar, and may also have overrated the strength of the Miraelesuit mark, these factors only support our affirmance of the ultimate conclusion that the marks are not likely to be directly confused.

Although a number of factors favored A & H, these do not upset the balance. In sum, we find no error in the District Court’s careful consideration and weighing of the factors, and its use of them as tools in reaching its ultimate finding of fact. We therefore will affirm the judgment with respect to the direct confusion claim.

V. The Reverse Confusion Claim

A. Introduction

We recently recognized the doctrine of “reverse confusion” as a distinct *228basis for a claim under § 43(a) of the Lanham Act. See Fisons, 30 F.3d at 475. While the essence of a direct confusion claim is that a junior user of a mark is said to free-ride on the “reputation and good will of the senior user by adopting a similar or identical mark,” id., reverse confusion occurs when “the junior user saturates the market with a similar trademark and overwhelms the senior user.” Id. (quoting Ameritech, Inc. v. American Info. Techs. Corp., 811 F.2d 960, 964 (6th Cir. 1987)). The harm flowing from reverse confusion is that

[t]he public comes to assume the senior user’s products are really the junior user’s or that the former has become somehow connected to the latter.... [T]he senior user loses the value of the trademark — its product identity, corporate identity, control over its goodwill and reputation, and ability to move into new markets.

Ameritech, Inc., 811 F.2d at 964; see also Fisons, 30 F.3d at 479; Sands, Taylor & Wood Co. v. Quaker Oats Co., 978 F.2d 947, 957 (7th Cir.1992); Banff, Ltd. v. Federated Dep’t Stores, Inc., 841 F.2d 486, 490-91 (2d Cir.1988); Capital Films Corp. v. Charles Fries Prods, Inc., 628 F.2d 387, 393 (5th Cir.1980); Big O Tire Dealers, Inc., v. Goodyear Tire & Rubber Co., 561 F.2d 1365, 1372 (10th Cir.1977). As we explained in Fisons, reverse confusion protects “smaller senior users ... against larger, more powerful companies who want to use identical or confusingly similar trademarks.” 30 F.3d at 475. Absent reverse confusion, “a company with a well established trade name and with the economic power to advertise extensively [would be immunized from suit] for a product name taken from a competitor.” Big O Tire Dealers, Inc., 561 F.2d at 1372 (citation omitted).The doctrine of reverse confusion — or, at least, some of its applications — is not without its critics. See, e.g., Thad G. Long & Alfred M. Marks, Reverse Confusion: Fundamentals and Limits, 84 Trademark Rep. 1, 2-3 (1994); Daniel D. Domenico, Note, Mark Madness: How Brent Musburger and the Miracle Bra May Have Led to a More Equitable and Efficient Understanding of the Reverse Confusion Doctrine in Trademark Law, 86 Va.L.Rev. 597, 613-14, 621-24 (2000).18 The chief danger inherent in recognizing reverse confusion claims is that innovative junior users, who have invested heavily in promoting a particular mark, will suddenly find their use of the mark blocked by plaintiffs who have not invested in, or promoted, their own marks. See Weiner King, Inc. v. Wiener King Corp., 615 F.2d 512, 522 (C.C.P.A.1980). Further, an overly-vigorous use of the doctrine of reverse confusion could potentially inhibit larger companies with established marks from expanding their product lines — for instance, had Victoria’s Secret thought, at the outset, that it would not be permitted carry over its popular The Miracle Bra mark from lingerie to swimwear, it might have chosen not to enter the swimsuit market at all.

This would be an undesirable result; in fact, it is precisely to allow a certain amount of “space” for companies to expand their product lines under established marks that we allow infringement suits against suppliers of noncompeting goods. See Interpace Corp. v. Lapp, Inc., 721 F.2d 460, 464 (3d Cir.1983). This is not to say that the reverse confusion doctrine does not have its proper place; as has been recognized, without the existence of such a claim, smaller business owners might not have any incentive to invest in their marks at all, for fear the mark could be usurped at will by a larger competitor. See SK & F, Co. v. Premo Pharm. Labs., Inc., 625 F.2d 1055, 1067 (3d Cir.1980) (“[P]ermitting piracy of ... identifying trade dress can only discourage other *229manufacturers from making a similar individual promotional effort”). However, these concerns do sensitize us to the potential untoward effects of an overenthusiastic enforcement of reverse confusion claims, although they cannot supersede our judicial recognition of the doctrine.

B. The Test for Reverse Confusion

As in a direct confusion claim, the ultimate question in a reverse confusion claim is whether there is a likelihood of consumer confusion as to the source or sponsorship of a product. See Fisons, 30 F.3d at 475. Although it would seem, somewhat counterintuitive to posit that the likelihood of confusion analysis changes from the direct confusion to the reverse confusion context,19 there are differences between the two situations that bear mentioning. Therefore, to clarify the test for reverse confusion that has developed in our jurisprudence, we will walk through the factors that a district court should consider (where relevant) in assessing a such a claim.

1. The Factors that are the Same

As an initial matter, there are several factors that should generally be analyzed in the same way for a reverse confusion claim as they are for a direct confusion claim.20 First, the attentiveness of consumers does not change (factor (3)); in both direct and reverse confusion, the question is whether this is the kind of product that consumers will care enough about to notice the differences, or purchase hastily with only a limited impression. See Fisons, 30 F.3d at 476 n. 12 (considering this factor in the same manner as it would for direct confusion). Second, and similarly, the degree to which the channels of trade and advertisement overlap (factor (7)) should be analyzed in the same fashion. See id. at 475-76 (analyzing the channels of trade in the same manner). Finally, Lapp factors (8) and (9), considering the similarity of the targets of the parties’ sales efforts and the similarity of products, are also analyzed no differently in the reverse confusion context. See id. at 475, 481 (treating these factors in the same way for reverse confusion as they would have been treated for direct confusion).

2. Similarity of the Marks

Generally speaking, the similarity ’ of the marks themselves is necessarily analyzed in the same way in direct and reverse confusion claims; the court looks to sight, sound, and meaning, and compares whether these elements combine to create a general commercial impression that is the same for the two marks. See, e.g., Fisons, 30 F.3d at 478-79 (analyzing the commercial impression of the marks in light of direct confusion principles). Therefore, a district court would not need to examine these in a different manner than it would in a direct confusion claim.

On the other hand, the direct confusion claim in this case was rejected by the District Court in considerable measure because the court felt that the Victoria’s Secret housemark, coupled with the disclaimer, alleviated any confusion that might otherwise result. See A & H IV, 57 F.Supp.2d at 168-69. Yet in the reverse confusion context, the presence of house-marks or disclaimers must obviously be treated differently than in the direct confu*230sion context. It is the essence of the reverse confusion claim that, when consumers come across the Miraclesuit in the stream of commerce, they will confuse it with The Miracle Bra and think that it is a Victoria’s Secret product. Therefore, the weight of a disclaimer on the Victoria’s Secret product is necessarily lessened. Because A & H puts no disclaimer on its product to distinguish it from The Miracle Bra, the consumer considering a purchase of the Miraclesuit will not have the same handy reminder that Miraclesuit is not associated with The Miracle Bra or Victoria’s Secret. This is not to say that such a disclaimer may not, in fact, mitigate confusion in some cases; if consumers are faced with the disclaimer every time they flip through the Victoria’s Secret catalogue, they are less likely to forget that Miracle-suit is unrelated to The Miracle Bra swimwear.

As to the presence of the housemark on the Victoria’s Secret product, not only is there the possibility that consumers will fail to remember the mark when encountering A & H’s swimwear, but there is also the possibility that the mark will aggravate, rather than mitigate, reverse confusion, by reinforcing the association of the word “miracle” exclusively with Victoria’s Secret. See, e.g., Sands, Taylor & Wood Co. v. Quaker Oats Co., 978 F.2d 947, 960 (7th Cir.1992); Americana Trading Inc. v. Russ Berrie & Co., 966 F.2d 1284, 1288 (9th Cir.1992). Of course, we do not suggest that this actually occurred in this particular ease; after all, the District Court observed that A & H typically includes its own housemark on Miraclesuits, see A & H IV, 57 F.Supp.2d at 160, but, because the court only conducted a likelihood of confusion analysis for the direct confusion claim, it only briefly addressed the significance of the A & H housemark, see id. at 168 n. 17.

Clearly, the proper significance to be accorded these facts is a matter best suited for the determination of the trial court. Instead, we merely highlight the questions raised by the use of the housemarks and disclaimers in order to emphasize that a district court must separately examine the similarity factor to determine whether there are any aspects of the analysis that should be different for a reverse confusion claim, and, if so, alter its examination accordingly.

3. Strength of the Marks

An important difference between reverse and direct confusion manifests in the analysis of the strength of the marks. As we explained supra, this factor requires consideration both of the mark’s commercial and conceptual strength. For ease of understanding, we will explain the appropriate treatment of commercial strength first, and the treatment of conceptual strength second.

a. Commercial Strength

It has been observed that a consumer first encountering a mark with one set of goods is likely to continue to associate the mark with those goods, and whether any subsequent confusion is “direct” or “reverse” will depend on whether the consumer’s first experience was with the junior or the senior user of the mark. See Banff, Ltd. v. Federated Dep’t Stores, Inc., 841 F.2d 486, 490 (2d Cir.1988) (acknowledging such a possibility); Long & Marks, supra, at 5. The greater the commercial disparity between the manufacturers, the more likely it is that a consumer’s first experience with a mark will be with one particular manufacturer. That is, if one manufacturer — junior or senior — expends tremendous sums in advertising while the other does not, consumers will be more likely to encounter the heavily advertised mark first. Where the greater advertising originates from the senior user, we are more likely to see a case of direct confusion; if the greater advertising originates from the junior user, reverse confusion is more likely. See 3 McCarthy, supra, § 23:10, at 23-32; cf. Fisons, 30 F.3d at 479 (observing that direct confusion involves a junior user *231“trad[ing] on” a senior user’s name and thus expending less on advertising, whereas reverse confusion involves the opposite pattern).

Logically, then, in a direct confusion claim, a plaintiff with a commercially strong mark is more likely to prevail than a plaintiff with a commercially weak mark. Conversely, in a reverse confusion claim, a plaintiff with a commercially weak mark is more likely to prevail than a plaintiff with a stronger mark, and this is particularly true when the plaintiffs weaker mark is pitted against a defendant with a far stronger mark. McCarthy has written that “the relatively large advertising and promotion of the junior user ... is the hallmark of a reverse confusion case.” 3 McCarthy, supra, § 23:10, at 23-37. “[T]he lack of commercial strength of the smaller senior user’s mark is to be given less weight in the analysis because it is the strength of the larger, junior user’s mark which results in reverse confusion.” Commerce Nat’l Ins. Servs., Inc. v. Commerce Ins. Agency, Inc., 214 F.3d 432, 444 (3d Cir.2000). As we explained in Fisons, “the evidence of commercial strength is different from what we expect in a case of forward confusion, where the junior user tries to palm off his goods as those of the senior user.” 30 F.3d at 479.

Therefore, in a reverse confusion claim, a court should analyze the “commercial strength” factor in terms of (1) the commercial strength of the junior user as compared to the senior user; and (2) any advertising or marketing campaign by the junior user that has resulted in a saturation in the public awareness of the junior user’s mark. See Fisons, 30 F.3d at 474, 479.

b. Distinctiveness or Conceptual Strength

In Fisons we remanded the case for the district court to “reevaluate[ ] ... distinctiveness as well as [the mark’s] commercial strength” for the reverse confusion claim. Fisons, 30 F.3d at 479. Although we explained that the evaluation of commercial strength would have to be altered for reverse confusion claims, we did not discuss how distinctiveness, or conceptual strength, should be re-weighed in light of our adoption of the reverse confusion doctrine. Nor did we clarify this aspect of our jurisprudence in Commerce National Insurance Services, Inc. v. Commerce Insurance Agency, Inc., 214 F.3d 432 (3d Cir.2000), where we referred to the different test for “commercial strength,” in a reverse confusion context, without reference to “conceptual strength.”

As stated above, in the paradigmatic reverse confusion case, the senior user has a commercially weak mark when compared with the junior user’s commercially strong mark. When it comes to conceptual strength, however, we believe that, just as in direct confusion cases, a strong mark should weigh in favor of a senior user. Our decision is supported by the fact that those courts that have clearly distinguished conceptual from commercial strength in the reverse confusion context have weighed a conceptually strong mark in the senior user’s favor, in the same manner as they would in direct confusion cases. See, e.g., Worthington Foods, Inc. v. Kellogg Co., 732 F.Supp. 1417, 1456 (S.D.Ohio 1990).

In H. Lubovsky, Inc. v. Esprit de Corp., 627 F.Supp. 483 (S.D.N.Y.1986), the court explained that conceptual distinctiveness was relevant in the same way for a reverse confusion claim because “if a customer saw a doll in a toy store bearing a strong familiar trademark like ‘Exxon,’ he might well assume that the oil company had gone into the toy business; if, on the other hand, he saw a doll bearing a familiar but weak laudatory trademark like Merit, he would be unlikely to assume that it is connected with the similarly named gasoline or cigarettes.” Id. at 487; see also Long & Marks, supra, at 22.

The H. Lubovsky logic resonates, for it makes more sense to hold that con*232ceptual strength, unlike commercial strength, works in the plaintiffs favor. That is, if we were to apply the rule stated above for commercial strength, i.e., weighing weakness in the plaintiffs favor, we would bring about the perverse result that less imaginative marks would be more likely to win reverse confusion claims than arbitrary or fanciful ones. We therefore hold that, as in direct confusion claims, a district court should weigh a conceptually strong mark in the plaintiffs favor, particularly when the mark is of such a distinctive character that, coupled with the relative similarity of the plaintiffs and defendant’s marks, a consumer viewing the plaintiffs product is likely to assume that such a mark would only have been adopted by a single source — i.e., the defendant.

4. The Intent of the Defendant

In the direct confusion context, the intent of the defendant is relevant to the extent that it bears on the likelihood of confusion analysis. As we have said:

In the likelihood of confusion inquiry ... we do not focus on a defendant’s bare intent to adopt a mark ... substantially identical to a plaintiffs mark ..., since there is little basis in fact or logic for supposing from a defendant’s intent to copy without more that the defendant’s actions will in fact result in confusion. Thus, what we have held is that a defendant’s intent to confuse or deceive consumers as to the product’s source may be highly probative of likelihood of confusion.

Versa Prods. Co., Inc. v. Bifold Co. (Mfg.) Ltd., 50 F.3d 189, 205 (3d Cir.1995) (emphasis omitted).

When reverse, rather than direct, confusion is alleged, “intent to confuse” is unlikely to be present. Cf. Fisons, 30 F.3d at 480. However, though perhaps unusual, should an intent to confuse exist, it would be relevant to the likelihood of confusion analysis in the same manner as it would for a direct confusion claim. For instance, in Commerce National Insurance Services, Inc. v. Commerce Insurance Agency, Inc., 214 F.3d 432 (3d Cir.2000), we were confronted with a situation in which the litigants had used very similar marks in noncompetitive industries for a number of years, each fully aware of the other and with no incidents of actual confusion. Eventually, however, the larger company expanded into the smaller company’s line of business, deliberately choosing to promote its services under an almost identical mark. In holding that the smaller company could maintain its claim against the larger for reverse confusion, we specifically highlighted the possibility that the larger company had adopted the mark with the deliberate intent of pushing its rival out of the market, and that it was this sort of usurpation of business identity that the reverse confusion doctrine was designed to prevent. See id. at 445.

As we have noted in our two prior cases on this issue, the defendant’s intent may be discovered through such inquiries as whether the defendant was aware of the senior user’s mark when it adopted its own mark, and whether the defendant considered that its adoption of the mark might result in confusion. See id. at 444; Fisons, 30 F.3d at 480. If such an intent to confuse does, in fact, exist in a reverse confusion case, it should weigh against the defendant in the same manner as it would in a direct confusion case. Cf. W.W.W. Pharm. Co., Inc. v. Gillette Co., 984 F.2d 567, 575 (2d Cir.1993) (weighing the “intent” factor in a reverse confusion case in the defendant’s favor because the plaintiff had not demonstrated an intent to confuse).

Although we recognize that our opinion in Fisons perhaps implied that mere carelessness, as opposed to deliberate intent to confuse, would weigh in a plaintiffs favor in a reverse confusion case, we are reluctant to adopt such an interpretation, as it would be manifestly out of step with our prior holdings regarding the rele-*233vanee of “intent” in trademark infringement claims. Cf. O. Hommel Co. v. Ferro Corp., 659 F.2d 340, 854 (3d Cir.1981). Ultimately, all of the Lapp factors are meant only to determine whether confusion is likely; mere carelessness, like deliberate copying, does not shed any light on this inquiry. Further, to the extent that the intent inquiry in the likelihood of confusion analysis carries with it the attribution of fault, there is no reason to ascribe higher penalties to a lower degree of fault because a particular case involves reverse, rather than direct, confusion. Finally, in light of the policy concerns implicated by the reverse confusion doctrine, it would be troubling indeed to hold that a lesser degree of culpability would weigh in the plaintiffs favor for a reverse confusion claim than it would for a direct confusion claim.

5. Factors Relating to Actual Confusion

As a matter of intuition, one would expect that in a reverse confusion claim, evidence of actual confusion would be as important as in a direct confusion claim, though the nature of the confusion that would be probative would be quite different. See Lang v. Retirement Living Publ’g Co., 949 F.2d 576, 583 (2d Cir.1991) (holding that evidence of “actual confusion” in which the public thought the senior user was the origin of the junior user’s products was irrelevant for a reverse confusion claim). As applied to this case, for example, evidence that consumers thought that The Miracle Bra was an A & H product would be probative on a direct confusion claim, but not on a reverse confusion claim. Conversely, evidence that consumers thought that Miraclesuit was a Victoria’s Secret product would support a reverse confusion claim, but not a direct confusion claim. This was apparently the District Court’s intuition; although it declined to consider A & H’s reverse confusion claim, it did observe that most of the evidence A & H had put forth with regard to “actual confusion” related to direct, rather than reverse, confusion. See A & H IV, 57 F.Supp.2d at 178 n. 32.

However, marshalling evidence of actual confusion is often difficult. See, e.g., Liquid Glass Enters., Inc. v. Dr. Ing. h.c.F. Porsche AG, 8 F.Supp.2d 398, 403 (D.N.J. 1998). In our view, if we were to create a rigid division between “direct” and “reverse” confusion evidence, we would run the risk of denying recovery to meritorious plaintiffs. For example, if a plaintiff alleged theories of both direct and reverse confusion and was able to prove a few instances of “actual” confusion in each direction, we might conclude that the plaintiff did not have enough evidence of either type to succeed on either of its claims, even though, taken together, all of the evidence of actual confusion would be probative of a real problem. As we explained in Part V.B.3, supra, the manifestation of consumer confusion as “direct” or “reverse” may merely be a function of the context in which the consumer first encountered the mark. Isolated instances of “direct” confusion may occur in a reverse confusion case, and vice-versa. See Long & Marks, supra, at 5. Though we might expect that, in most instances, the consumer’s first encounter will be with the mark that has greater commercial strength, this will not invariably be the case.

Given the problems litigants typically encounter in locating evidence of actual confusion, then, we decline to create a strict bar to the use of “direct” confusion evidence in a “reverse” confusion case, or vice versa. However, evidence working in the same direction as the claim is preferred, and “misfitting” evidence must be treated carefully, for large amounts of one type of confusion in a claim for a different type may in fact work against the plaintiff. For instance, the existence of reverse confusion might disprove a plaintiffs claim that its descriptive mark has secondary meaning, thus resulting in no recovery at all. See Jefferson Home Furniture Co., *234 Inc. v. Jefferson Furniture Co., Inc., 349 So.2d 5, 8 (Ala.1977).

It follows that the other factor relating to actual confusion, Lapp factor (4), examining the time the mark has been used without evidence of actual confusion, should be approached similarly.

6. Other Relevant Facts

The final factor of the Lapp test directs courts to look at “other facts suggesting that the consuming public might expect the prior owner to manufacture both products, or expect the prior owner to manufacture a product in the defendant’s market, or expect that the prior owner is likely to expand into the defendant’s market.” This factor is necessarily transformed in the reverse confusion context to an examination of other facts suggesting that the consuming public might expect the larger, more powerful company to manufacture both products, or expect the larger company to manufacture a product in the plaintiffs market, or expect that the larger company is likely to expand into the plaintiffs market. See Fisons, 30 F.3d at 480 (directing the district court to examine facts suggesting that the public might think that the junior user would expand into the senior user’s market).

7. Summary of the Test for Reverse Confusion

In sum, in the typical case in which there is a claim of reverse confusion, a court should examine the following factors as aids in its determination whether or not there is a likelihood of such confusion:

(1) the degree of similarity between the owner’s mark and the alleged infringing mark;

(2) the strength of the two marks, weighing both a commercially strong junior user’s márk and a conceptually strong senior user’s mark in the senior user’s favor;

(3) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;

(4) the length of time the defendant has used the mark without evidence of actual confusion arising;

(5) the intent of the defendant in adopting the mark;

(6) the evidence of actual confusion;

(7) whether the goods, competing or not competing, are marketed through the same channels of trade and advertised through the same media;

(8) the extent to which the targets of the parties’ sales efforts are the same;

(9) the relationship of the goods in the minds of consumers, whether because of the near-identity of the products, the similarity of function, or other factors;

(10) other facts suggesting that the consuming public might expect the larger, more powerful company to manufacture both products, or expect the larger company to manufacture a product in the plaintiffs market, or expect that the larger company is likely to expand into the plaintiffs market.

As with the test for direct confusion, no one factor is dispositive, and in individual cases, particular factors may not be probative on the issue of likelihood of confusion. “The weight given to each factor in the overall picture, as well as its weighing for plaintiff or defendant, must be done on an individual fact-specific basis.” Fisons, 30 F.3d at 476 n. 11.

C. The District Court’s Opinion

The District Court approached the reverse confusion claim in a different manner from that described in the foregoing section. It held that before engaging the reverse confusion factors, A & H needed to demonstrate, as a threshold matter, that Victoria’s Secret “used their economic power to overwhelm the market with advertising” of their product. A & H IV, 57 F.Supp.2d at 178. The opinion therefore *235focused only on a comparison between the commercial strengths of the Miraclesuit and The Miracle Bra, i.e., only on one aspect of the “mark strength” inquiry. The District Court found that Victoria’s Secret had saturated the market with $13 million in The Miracle Bra advertising, and that a “meaningful portion” of the advertising went towards promoting The Miracle Bra swimwear. It then compared Victoria’s Secret’s effort to A & H’s effort to promote Miraclesuit swimwear; A & H spent over $1.2 million on advertising, and received $1.5 million in free publicity.21

The court noted that A & H did not dispute — even heralded — the fact that its campaign was widely successful. Therefore, it found that “[plaintiffs’ [sic] are not entirely without market power in the swimwear industry” and that “[i]n light of Plaintiffs’ advertising campaign, we find that, in comparison, Defendants did not saturate the marketplace with its advertising to promote The Miracle Bra swimwear.” Id. at 177. Consequently, the District Court concluded that the doctrine of reverse confusion was not even implicated, and “decline[d] to examine whether a likelihood of reverse confusion exists.” Id. at 178.

In A & H III, we began our discussion of reverse confusion with an explanation of the phenomenon: “Reverse confusion occurs when a larger more powerful company uses the trademark of a smaller, less powerful senior owner and thereby causes likely confusion as to the source of the senior user’s goods or services.” 166 F.3d at 207 (quoting Fisons, 30 F.3d at 474.). In Fisons, we stated that it would be necessary to recognize the reverse confusion doctrine to provide protection to “smaller, senior users” against “larger, more powerful companies.” 30 F.3d at 475. While these are accurate statements about the doctrine of reverse confusion, the District Court appears to have interpreted them as establishing a separate, threshold step that must be examined pri- or to engaging the Lapp test, and, finding the threshold not met, concluded that A & H’s power to advertise extensively precluded it from bringing a reverse confusion claim. The District Court was correct to note that commercial disparity is, in fact, a factor to consider (factor (2), to be specific); however, it erred in applying a threshold commercial disparity requirement, in effect making this sole factor determinative and treating the reverse confusion inquiry as requiring a two-step process.

The quoted statements are understandably confusing, and the method of applying the doctrine of reverse confusion is admittedly still developing. However, a close examination of Fisons, from which the quotes hail, reveals that they are nothing more than descriptions of the phenomenon of reverse confusion, and do not establish an initial inquiry that a court needs to make in order to apply the reverse confusion analysis. In Fisons we held that the Lapp factors constitute the method courts should use in order to determine if a likelihood of reverse confusion exists. We made no mention of a threshold requirement, nor did we direct the District Court, on remand, to use one.

The District Court further explained its choice not to apply the reverse confusion factors by reference to the fact that several reverse confusion cases involve enormous junior companies pitted against tiny senior companies; it cited these cases and concluded that the same degree of economic disparity was non-existent in this case. See A & H IV, 57 F.Supp.2d at 178. The difficulty with this conclusion is that in Fisons itself, in which we found a viable reverse confusion claim, Fisons had fully 25% of the peat moss market, greater than A & H’s 10% of the swimwear market. This suggests that a company need not be all that weak within its market in order to *236bring a viable reverse confusion claim. See Fisons, 30 F.3d at 479; see also Fuji Photo Film v. Shinohara Shoji Kabushiki, 754 F.2d 591 (5th Cir.1985) (finding a viable reverse confusion claim where the plaintiff had spent millions of dollars). In fact, in stating that “[i]n reverse confusion, the junior user is typically a wealthier, more powerful company who can overwhelm the market with advertising,” 30 F.3d at 479 (emphasis added), we implied that there might be the rare case in which reverse confusion exists where the junior company overwhelms the senior user with advertisements although it is not wealthier and more powerful.

In short, we hold that, although economic disparity between the companies and the marks is an important consideration in the evaluation of the marks’ commercial strength, the District Court legally erred in fashioning a threshold “economic disparity” requirement before a reverse confusion claim will even be considered. Because the District Court failed to undertake the Lapp analysis with respect to A & H’s reverse confusion claim, we must vacate the judgment with respect to that claim and remand to the District Court for a redetermination of those factors that receive different treatment under direct and reverse confusion theories, and for a reweighing of all of the factors once those redeterminations have been made.

D. Guidance for Remand

1.Introduction

If the record suggested that, under the test we have set forth, A & H could not succeed as a matter of law in a reverse confusion claim, we would be bound to explicate our reasoning and affirm the judgment of the District Court. However, we do not so conclude, and hence vacatur and remand is necessary. Inasmuch as we have clarified the law of reverse confusion in this circuit by filling the gaps left in Fisons, it will be useful to the District Court if we comment on the extent to which it needs to revisit the various issues. As explained supra, the factors concerning the market, sales, and function similarity (factors (3), (7), (8) & (9)) need not be reexamined for the reverse confusion claim because the District Court has already discussed them in connection with direct confusion and there is typically no difference in the analysis of these factors for reverse and direct confusion claims. As for the “actual confusion” factors ((4) & (6)), the District Court did not credit the evidence proffered by A & H and, in its best light, regarded it as de minimis. A & H’s evidence of actual confusion primarily supported its claim for direct, rather than reverse, confusion, see supra Section IV.G, with the only exception being the Women’s Wear Daily article mentioning “the introduction of the Miracle Swimsuit in the upcoming Victoria’s Secret catalog.” Because these incidents, though relevant to reverse confusion, are more probative of direct confusion but the District Court felt that they were too weak to support even that claim, the court may, but need not, reexamine this factor on remand.

2. Similarity of the Marks

When addressing the direct confusion claim, the District Court placed great weight on the presence of Victoria’s Secret’s housemark and disclaimer when it concluded that the marks were not confusingly similar. See A & H IV, 57 F.Supp.2d at 168-69. However, as we explained supra, although such embellishments of the junior user’s mark may still have relevance in the reverse confusion context, their weight must necessarily be reevaluated. Therefore, on remand, the District Court should reconsider the similarity of the marks in light of A & H’s reverse confusion claims.

3. Strength of the Marks

The District Court did not consider the commercial strength of the marks within the ambit of the reverse confusion Fisons analysis, but it functionally did as *237much in its “threshold” determination that A & H lacked sufficient “economic disparity” relative to Victoria’s Secret to advance a reverse confusion claim. Therefore, the District Court essentially demonstrated that it weighed this factor in favor of Victoria’s Secret. However, in comparing the relative commercial strengths of the products, the District Court committed clear error: Although it considered the free publicity received by A & H in determining its commercial strength, it did not consider the free publicity received by Victoria’s Secret. This led to an inaccurate comparison of their relative commercial vitality. Had the court used the same calipers to measure the commercial strength of each, it might have determined that the Miraclesuit had less commercial strength relative to The Miracle Bra. Of course, the court might well deem the difference unimportant, but we cannot say that either Victoria’s Secret or A & H should have this factor weighed in its favor as a matter of law.

Furthermore, the court should have also considered the conceptual strength of the Miraclesuit mark, according to the standards for conceptual strength set forth in Sections IV.B and V.B.3, supra. In so doing, the court must gauge the strength of the Miraclesuit mark and must consider whether — as with the example set forth supra involving a doll with the mark “Exxon” — the Miracle Bra/Miraclesuit marks are so distinctive that, when considered simultaneously with the court’s determination as to their similarities, consumers with a general awareness of The Miracle Bra swimsuit are likely to assume that the Miraclesuit is a Victoria’s Secret product.

4. Intent

In its evaluation of A & H’s direct confusion claim, the District Court concluded that Victoria’s Secret’s “choice to extend The Miracle Bra mark to swimwear was for legitimate reasons, rather than out of bad faith.” A & H IV, 57 F.Supp.2d at 172. However, because it “decline[d] to examine whether a likelihood of reverse confusion exists,” id. at 178, the District Court focused on whether Victoria’s Secret had intended to “profit[ ] from the notoriety of Plaintiffs’ Miraclesuit mark,” A & H TV, 57 F.Supp.2d at 173-74. The court did not specifically address the question whether Victoria’s Secret, rather than intending to “free ride” on A & H’s goodwill, instead intended to usurp it by deliberately undertaking to cause consumer confusion (and thereby destroy A & H’s business identity). On remand, the District Court should consider whether its previous finding of Victoria’s Secret’s good faith is dispositive of the reverse confusion intent analysis, or whether further examination of this issue is warranted.

E. Summary

Although we believe that the District Court’s evaluation of individual factors relating to market, sales, and functionality would have remained unchanged had it examined A & H’s reverse confusion claim in light of the Lapp factors, we simply do not know how it would have treated the commercial strength and mark similarity factors had it considered the free advertising The Miracle Bra received, or the effect of the housemark and disclaimer in the reverse confusion context. We also cannot predict what the result would have been had the District Court examined the “intent” factor in light of A & H’s reverse confusion claim, or how the District Court would have weighed the various factors had it not determined that there was a threshold commercial disparity requirement.

The question of likelihood of confusion is ultimately one of fact, and we cannot roll up our sleeves and engage in the balancing ourselves. In its balancing on the direct confusion claim, the District Court found that the case was close, holding that no likelihood of confusion existed “solely based on the presumption that Defendants will continue to use the disclaimer when marketing The Miracle Bra swimwear,” A *238 & HIV, 57 F.Supp.2d at 169. As we have explained, we believe that Victoria’s Secret’s disclaimer has a lessened significance for reverse confusion. We also believe that the conceptual strength of the Miraclesuit mark must be reevaluated. Under these circumstances, we cannot say as a matter of law that a different weighing of the factors could not have influenced the District Court to make a different finding of ultimate fact, thus necessitating a remand.

VI. Conclusion

We will affirm the District Court’s judgment for Victoria’s Secret on the direct confusion claim. However, we will vacate the judgment with respect to the reverse confusion claim, and remand to the District Court for further proceedings consistent with this opinion. The District Court may wish to hear and consider additional evidence from the parties on reverse confusion. The parties shall bear their own costs.

Chief Justice ROBERTS delivered the opinion of the Court.

A United States patent entitles the patent holder (the "patentee"), for a period of 20 years, to "exclude others from making, using, offering for sale, or selling [its] invention throughout the United States or importing the invention into the United States." 35 U.S.C. § 154(a). Whoever engages in one of these acts "without authority" from the patentee may face liability for patent infringement. § 271(a).

When a patentee sells one of its products, however, the patentee can no longer control that item through the patent laws-its patent rights are said to "exhaust." The purchaser and all subsequent owners are free to use or resell the product just like any other item of personal property, without fear of an infringement lawsuit.

This case presents two questions about the scope of the patent exhaustion doctrine: First, whether a patentee that sells an item under an express restriction on the purchaser's right to reuse or resell the product may enforce that restriction through an infringement lawsuit. And second, whether a patentee exhausts its patent rights by selling its product outside the United States, where American patent laws do not apply. We conclude that a patentee's decision to sell a product exhausts all of its patent rights in that item, regardless of any restrictions the patentee purports to impose or the location of the sale.

I

The underlying dispute in this case is about laser printers-or, more specifically, the cartridges that contain the powdery substance, known as toner, that laser printers use to make an image appear on paper. Respondent Lexmark International, Inc. designs, manufactures, and sells toner cartridges to consumers in the United States and around the globe. It owns a number of patents that cover components of those cartridges and the manner in which they are used.

When toner cartridges run out of toner they can be refilled and used again. This creates an opportunity for other companies-known as remanufacturers-to acquire empty Lexmark cartridges from purchasers in the United States and abroad, refill them with toner, and then resell them at a lower price than the new ones Lexmark puts on the shelves.

Not blind to this business problem, Lexmark structures its sales in a way that encourages customers to return spent cartridges. It gives purchasers two options: One is to buy a toner cartridge at full price, with no strings attached. The other is to buy a cartridge at roughly 20-percent *1530off through Lexmark's "Return Program." A customer who buys through the Return Program still owns the cartridge but, in exchange for the lower price, signs a contract agreeing to use it only once and to refrain from transferring the empty cartridge to anyone but Lexmark. To enforce this single-use/no-resale restriction, Lexmark installs a microchip on each Return Program cartridge that prevents reuse once the toner in the cartridge runs out.

Lexmark's strategy just spurred remanufacturers to get more creative. Many kept acquiring empty Return Program cartridges and developed methods to counteract the effect of the microchips. With that technological obstacle out of the way, there was little to prevent the remanufacturers from using the Return Program cartridges in their resale business. After all, Lexmark's contractual single-use/no-resale agreements were with the initial customers, not with downstream purchasers like the remanufacturers.

Lexmark, however, was not so ready to concede that its plan had been foiled. In 2010, it sued a number of remanufacturers, including petitioner Impression Products, Inc., for patent infringement with respect to two groups of cartridges. One group consists of Return Program cartridges that Lexmark sold within the United States. Lexmark argued that, because it expressly prohibited reuse and resale of these cartridges, the remanufacturers infringed the Lexmark patents when they refurbished and resold them. The other group consists of all toner cartridges that Lexmark sold abroad and that remanufacturers imported into the country. Lexmark claimed that it never gave anyone authority to import these cartridges, so the remanufacturers ran afoul of its patent rights by doing just that.

Eventually, the lawsuit was whittled down to one defendant, Impression Products, and one defense: that Lexmark's sales, both in the United States and abroad, exhausted its patent rights in the cartridges, so Impression Products was free to refurbish and resell them, and to import them if acquired abroad. Impression Products filed separate motions to dismiss with respect to both groups of cartridges. The District Court granted the motion as to the domestic Return Program cartridges, but denied the motion as to the cartridges Lexmark sold abroad. Both parties appealed.

The Federal Circuit considered the appeals en banc and ruled for Lexmark with respect to both groups of cartridges. The court began with the Return Program cartridges that Lexmark sold in the United States. Relying on its decision in Mallinckrodt, Inc. v. Medipart, Inc., 976 F.2d 700 (1992), the Federal Circuit held that a patentee may sell an item and retain the right to enforce, through patent infringement lawsuits, "clearly communicated, ... lawful restriction[s] as to post-sale use or resale." 816 F.3d 721, 735 (2016). The exhaustion doctrine, the court reasoned, derives from the prohibition on making, using, selling, or importing items "without authority." Id., at 734 (quoting 35 U.S.C. § 271(a) ). When you purchase an item you presumptively also acquire the authority to use or resell the item freely, but that is just a presumption; the same authority does not run with the item when the seller restricts post-sale use or resale. 816 F.3d, at 742. Because the parties agreed that Impression Products knew about Lexmark's restrictions and that those restrictions did not violate any laws, the Federal Circuit concluded that Lexmark's sales had not exhausted all of its patent rights, and that the company could sue for infringement when Impression Products refurbished *1531and resold Return Program cartridges.

As for the cartridges that Lexmark sold abroad, the Federal Circuit once again looked to its precedent. In Jazz Photo Corp. v. International Trade Commission, 264 F.3d 1094 (2001), the court had held that a patentee's decision to sell a product abroad did not terminate its ability to bring an infringement suit against a buyer that "import[ed] the article and [sold] ... it in the United States." 816 F.3d, at 726-727. That rule, the court concluded, makes good sense: Exhaustion is justified when a patentee receives "the reward available from [selling in] American markets," which does not occur when the patentee sells overseas, where the American patent offers no protection and therefore cannot bolster the price of the patentee's goods. Id., at 760-761. As a result, Lexmark was free to exercise its patent rights to sue Impression Products for bringing the foreign-sold cartridges to market in the United States.

Judge Dyk, joined by Judge Hughes, dissented. In their view, selling the Return Program cartridges in the United States exhausted Lexmark's patent rights in those items because any "authorized sale of a patented article ... free[s] the article from any restrictions on use or sale based on the patent laws." Id., at 775-776. As for the foreign cartridges, the dissenters would have held that a sale abroad also results in exhaustion, unless the seller "explicitly reserve[s] [its] United States patent rights" at the time of sale. Id., at 774, 788. Because Lexmark failed to make such an express reservation, its foreign sales exhausted its patent rights.

We granted certiorari to consider the Federal Circuit's decisions with respect to both domestic and international exhaustion, 580 U.S. ----, 137 S.Ct. 546, 196 L.Ed.2d 442 (2016), and now reverse.

II

A

First up are the Return Program cartridges that Lexmark sold in the United States. We conclude that Lexmark exhausted its patent rights in these cartridges the moment it sold them. The single-use/no-resale restrictions in Lexmark's contracts with customers may have been clear and enforceable under contract law, but they do not entitle Lexmark to retain patent rights in an item that it has elected to sell.

The Patent Act grants patentees the "right to exclude others from making, using, offering for sale, or selling [their] invention[s]." 35 U.S.C. § 154(a). For over 160 years, the doctrine of patent exhaustion has imposed a limit on that right to exclude. See Bloomer v. McQuewan, 14 How. 539, 14 L.Ed. 532 (1853). The limit functions automatically: When a patentee chooses to sell an item, that product "is no longer within the limits of the monopoly" and instead becomes the "private, individual property" of the purchaser, with the rights and benefits that come along with ownership. Id., at 549-550. A patentee is free to set the price and negotiate contracts with purchasers, but may not, "by virtue of his patent, control the use or disposition" of the product after ownership passes to the purchaser. United States v. Univis Lens Co., 316 U.S. 241, 250, 62 S.Ct. 1088, 86 L.Ed. 1408 (1942) (emphasis added). The sale "terminates all patent rights to that item." Quanta Computer, Inc. v. LG Electronics, Inc., 553 U.S. 617, 625, 128 S.Ct. 2109, 170 L.Ed.2d 996 (2008).

This well-established exhaustion rule marks the point where patent rights yield to the common law principle against restraints on alienation. The Patent Act *1532"promote[s] the progress of science and the useful arts by granting to [inventors] a limited monopoly" that allows them to "secure the financial rewards" for their inventions. Univis, 316 U.S., at 250, 62 S.Ct. 1088. But once a patentee sells an item, it has "enjoyed all the rights secured" by that limited monopoly. Keeler v. Standard Folding Bed Co., 157 U.S. 659, 661, 15 S.Ct. 738, 39 L.Ed. 848 (1895). Because "the purpose of the patent law is fulfilled ... when the patentee has received his reward for the use of his invention," that law furnishes "no basis for restraining the use and enjoyment of the thing sold." Univis, 316 U.S., at 251, 62 S.Ct. 1088.

We have explained in the context of copyright law that exhaustion has "an impeccable historic pedigree," tracing its lineage back to the "common law's refusal to permit restraints on the alienation of chattels." Kirtsaeng v. John Wiley & Sons, Inc., 568 U.S. 519, 538, 133 S.Ct. 1351, 185 L.Ed.2d 392 (2013). As Lord Coke put it in the 17th century, if an owner restricts the resale or use of an item after selling it, that restriction "is voide, because ... it is against Trade and Traffique, and bargaining and contracting betweene man and man." 1 E. Coke, Institutes of the Laws of England § 360, p. 223 (1628); see J. Gray, Restraints on the Alienation of Property § 27, p. 18 (2d ed. 1895) ("A condition or conditional limitation on alienation attached to a transfer of the entire interest in personalty is as void as if attached to a fee simple in land").

This venerable principle is not, as the Federal Circuit dismissively viewed it, merely "one common-law jurisdiction's general judicial policy at one time toward anti-alienation restrictions." 816 F.3d, at 750. Congress enacted and has repeatedly revised the Patent Act against the backdrop of the hostility toward restraints on alienation. That enmity is reflected in the exhaustion doctrine. The patent laws do not include the right to "restrain [ ] ... further alienation" after an initial sale; such conditions have been "hateful to the law from Lord Coke's day to ours" and are "obnoxious to the public interest." Straus v. Victor Talking Machine Co., 243 U.S. 490, 501, 37 S.Ct. 412, 61 L.Ed. 866 (1917). "The inconvenience and annoyance to the public that an opposite conclusion would occasion are too obvious to require illustration." Keeler, 157 U.S., at 667, 15 S.Ct. 738.

But an illustration never hurts. Take a shop that restores and sells used cars. The business works because the shop can rest assured that, so long as those bringing in the cars own them, the shop is free to repair and resell those vehicles. That smooth flow of commerce would sputter if companies that make the thousands of parts that go into a vehicle could keep their patent rights after the first sale. Those companies might, for instance, restrict resale rights and sue the shop owner for patent infringement. And even if they refrained from imposing such restrictions, the very threat of patent liability would force the shop to invest in efforts to protect itself from hidden lawsuits. Either way, extending the patent rights beyond the first sale would clog the channels of commerce, with little benefit from the extra control that the patentees retain. And advances in technology, along with increasingly complex supply chains, magnify the problem. See Brief for Costco Wholesale Corp. et al. as Amici Curiae 7-9; Brief for Intel Corp. et al. as Amici Curiae 17, n. 5 ("A generic smartphone assembled from various high-tech components could practice an estimated 250,000 patents").

This Court accordingly has long held that, even when a patentee sells an item under an express restriction, the patentee does not retain patent rights in that *1533product. In Boston Store of Chicago v. American Graphophone Co., for example, a manufacturer sold graphophones-one of the earliest devices for recording and reproducing sounds-to retailers under contracts requiring those stores to resell at a specific price. 246 U.S. 8, 17-18, 38 S.Ct. 257, 62 L.Ed. 551 (1918). When the manufacturer brought a patent infringement suit against a retailer who sold for less, we concluded that there was "no room for controversy" about the result: By selling the item, the manufacturer placed it "beyond the confines of the patent law, [and] could not, by qualifying restrictions as to use, keep [it] under the patent monopoly." Id., at 20, 25, 38 S.Ct. 257.

Two decades later, we confronted a similar arrangement in United States v. Univis Lens Co . There, a company that made eyeglass lenses authorized an agent to sell its products to wholesalers and retailers only if they promised to market the lenses at fixed prices. The Government filed an antitrust lawsuit, and the company defended its arrangement on the ground that it was exercising authority under the Patent Act. We held that the initial sales "relinquish [ed] ... the patent monopoly with respect to the article[s] sold," so the "stipulation ... fixing resale prices derive[d] no support from the patent and must stand on the same footing" as restrictions on unpatented goods. 316 U.S., at 249-251, 62 S.Ct. 1088.

It is true that Boston Store and Univis involved resale price restrictions that, at the time of those decisions, violated the antitrust laws. But in both cases it was the sale of the items, rather than the illegality of the restrictions, that prevented the patentees from enforcing those resale price agreements through patent infringement suits. And if there were any lingering doubt that patent exhaustion applies even when a sale is subject to an express, otherwise lawful restriction, our recent decision in Quanta Computer, Inc. v. LG Electronics, Inc. settled the matter. In that case, a technology company-with authorization from the patentee-sold microprocessors under contracts requiring purchasers to use those processors with other parts that the company manufactured. One buyer disregarded the restriction, and the patentee sued for infringement. Without so much as mentioning the lawfulness of the contract, we held that the patentee could not bring an infringement suit because the "authorized sale ... took its products outside the scope of the patent monopoly." 553 U.S., at 638, 128 S.Ct. 2109.

Turning to the case at hand, we conclude that this well-settled line of precedent allows for only one answer: Lexmark cannot bring a patent infringement suit against Impression Products to enforce the single-use/no-resale provision accompanying its Return Program cartridges. Once sold, the Return Program cartridges passed outside of the patent monopoly, and whatever rights Lexmark retained are a matter of the contracts with its purchasers, not the patent law.

B

The Federal Circuit reached a different result largely because it got off on the wrong foot. The "exhaustion doctrine," the court believed, "must be understood as an interpretation of" the infringement statute, which prohibits anyone from using or selling a patented article "without authority" from the patentee. 816 F.3d, at 734 (quoting 35 U.S.C. § 271(a) ). Exhaustion reflects a default rule that a patentee's decision to sell an item "presumptively grant[s] 'authority' to the purchaser to use it and resell it." 816 F.3d, at 742. But, the Federal Circuit explained, the patentee does not have to hand over the full "bundle *1534of rights" every time. Id., at 741 (internal quotation marks omitted). If the patentee expressly withholds a stick from the bundle-perhaps by restricting the purchaser's resale rights-the buyer never acquires that withheld authority, and the patentee may continue to enforce its right to exclude that practice under the patent laws.

The misstep in this logic is that the exhaustion doctrine is not a presumption about the authority that comes along with a sale; it is instead a limit on "the scope of the patentee's rights ." United States v. General Elec. Co., 272 U.S. 476, 489, 47 S.Ct. 192, 71 L.Ed. 362 (1926) (emphasis added). The right to use, sell, or import an item exists independently of the Patent Act. What a patent adds-and grants exclusively to the patentee-is a limited right to prevent others from engaging in those practices. See Crown Die & Tool Co. v. Nye Tool & Machine Works, 261 U.S. 24, 35, 43 S.Ct. 254, 67 L.Ed. 516 (1923). Exhaustion extinguishes that exclusionary power. See Bloomer, 14 How., at 549 (the purchaser "exercises no rights created by the act of Congress, nor does he derive title to [the item] by virtue of the ... exclusive privilege granted to the patentee"). As a result, the sale transfers the right to use, sell, or import because those are the rights that come along with ownership, and the buyer is free and clear of an infringement lawsuit because there is no exclusionary right left to enforce.

The Federal Circuit also expressed concern that preventing patentees from reserving patent rights when they sell goods would create an artificial distinction between such sales and sales by licensees. Patentees, the court explained, often license others to make and sell their products, and may place restrictions on those licenses. A computer developer could, for instance, license a manufacturer to make its patented devices and sell them only for non-commercial use by individuals. If a licensee breaches the license by selling a computer for commercial use, the patentee can sue the licensee for infringement. And, in the Federal Circuit's view, our decision in General Talking PicturesCorp. v. Western Elec. Co., 304 U.S. 175, 58 S.Ct. 849, 82 L.Ed. 1273, aff'd on reh'g, 305 U.S. 124, 59 S.Ct. 116, 83 L.Ed. 81 (1938), established that-when a patentee grants a license "under clearly stated restrictions on post-sale activities" of those who purchase products from the licensee-the patentee can also sue for infringement those purchasers who knowingly violate the restrictions. 816 F.3d, at 743-744. If patentees can employ licenses to impose post-sale restrictions on purchasers that are enforceable through infringement suits, the court concluded, it would make little sense to prevent patentees from doing so when they sell directly to consumers.

The Federal Circuit's concern is misplaced. A patentee can impose restrictions on licensees because a license does not implicate the same concerns about restraints on alienation as a sale. Patent exhaustion reflects the principle that, when an item passes into commerce, it should not be shaded by a legal cloud on title as it moves through the marketplace. But a license is not about passing title to a product, it is about changing the contours of the patentee's monopoly: The patentee agrees not to exclude a licensee from making or selling the patented invention, expanding the club of authorized producers and sellers. See General Elec. Co., 272 U.S., at 489-490, 47 S.Ct. 192. Because the patentee is exchanging rights, not goods, it is free to relinquish only a portion of its bundle of patent protections.

A patentee's authority to limit licensees does not, as the Federal Circuit thought, mean that patentees can use licenses *1535to impose post-sale restrictions on purchasers that are enforceable through the patent laws. So long as a licensee complies with the license when selling an item, the patentee has, in effect, authorized the sale. That licensee's sale is treated, for purposes of patent exhaustion, as if the patentee made the sale itself. The result: The sale exhausts the patentee's rights in that item. See Hobbie v. Jennison, 149 U.S. 355, 362-363, 13 S.Ct. 879, 37 L.Ed. 766 (1893). A license may require the licensee to impose a restriction on purchasers, like the license limiting the computer manufacturer to selling for non-commercial use by individuals. But if the licensee does so-by, perhaps, having each customer sign a contract promising not to use the computers in business-the sale nonetheless exhausts all patent rights in the item sold. See Motion Picture Patents Co. v. Universal Film Mfg. Co., 243 U.S. 502, 506-507, 516, 37 S.Ct. 416, 61 L.Ed. 871 (1917). The purchasers might not comply with the restriction, but the only recourse for the licensee is through contract law, just as if the patentee itself sold the item with a restriction.

General Talking Pictures involved a fundamentally different situation: There, a licensee "knowingly ma[de] ... sales ... outside the scope of its license." 304 U.S., at 181-182, 58 S.Ct. 849 (emphasis added). We treated the sale "as if no license whatsoever had been granted" by the patentee, which meant that the patentee could sue both the licensee and the purchaser-who knew about the breach-for infringement. General Talking Pictures Corp. v. Western Elec. Co., 305 U.S. 124, 127, 59 S.Ct. 116, 83 L.Ed. 81 (1938). This does not mean that patentees can use licenses to impose post-sale restraints on purchasers. Quite the contrary: The licensee infringed the patentee's rights because it did not comply with the terms of its license, and the patentee could bring a patent suit against the purchaser only because the purchaser participated in the licensee's infringement. General Talking Pictures, then, stands for the modest principle that, if a patentee has not given authority for a licensee to make a sale, that sale cannot exhaust the patentee's rights.

In sum, patent exhaustion is uniform and automatic. Once a patentee decides to sell-whether on its own or through a licensee-that sale exhausts its patent rights, regardless of any post-sale restrictions the patentee purports to impose, either directly or through a license.

III

Our conclusion that Lexmark exhausted its patent rights when it sold the domestic Return Program cartridges goes only halfway to resolving this case. Lexmark also sold toner cartridges abroad and sued Impression Products for patent infringement for "importing [Lexmark's] invention into the United States." 35 U.S.C. § 154(a). Lexmark contends that it may sue for infringement with respect to all of the imported cartridges-not just those in the Return Program-because a foreign sale does not trigger patent exhaustion unless the patentee "expressly or implicitly transfer[s] or license[s]" its rights. Brief for Respondent 36-37. The Federal Circuit agreed, but we do not. An authorized sale outside the United States, just as one within the United States, exhausts all rights under the Patent Act.

This question about international exhaustion of intellectual property rights has also arisen in the context of copyright law. Under the "first sale doctrine," which is codified at 17 U.S.C. § 109(a), when a copyright owner sells a lawfully made copy of its work, it loses the power to restrict the purchaser's freedom "to sell or otherwise dispose of ... that copy." In *1536Kirtsaeng v. John Wiley & Sons, Inc., we held that this " 'first sale' [rule] applies to copies of a copyrighted work lawfully made [and sold] abroad." 568 U.S., at 525, 133 S.Ct. 1351. We began with the text of § 109(a), but it was not decisive: The language neither "restrict [s] the scope of [the] 'first sale' doctrine geographically," nor clearly embraces international exhaustion. Id., at 528-533, 133 S.Ct. 1351. What helped tip the scales for global exhaustion was the fact that the first sale doctrine originated in "the common law's refusal to permit restraints on the alienation of chattels." Id., at 538, 133 S.Ct. 1351. That "common-law doctrine makes no geographical distinctions." Id., at 539, 133 S.Ct. 1351. The lack of any textual basis for distinguishing between domestic and international sales meant that "a straightforward application" of the first sale doctrine required the conclusion that it applies overseas. Id., at 540, 133 S.Ct. 1351 (internal quotation marks omitted).

Applying patent exhaustion to foreign sales is just as straightforward. Patent exhaustion, too, has its roots in the antipathy toward restraints on alienation, see supra, at 1528 - 1533, and nothing in the text or history of the Patent Act shows that Congress intended to confine that borderless common law principle to domestic sales. In fact, Congress has not altered patent exhaustion at all; it remains an unwritten limit on the scope of the patentee's monopoly. See Astoria Fed. Sav. & Loan Assn. v. Solimino, 501 U.S. 104, 108, 111 S.Ct. 2166, 115 L.Ed.2d 96 (1991) ("[W]here a common-law principle is well established, ... courts may take it as given that Congress has legislated with an expectation that the principle will apply except when a statutory purpose to the contrary is evident" (internal quotation marks omitted)). And differentiating the patent exhaustion and copyright first sale doctrines would make little theoretical or practical sense: The two share a "strong similarity ... and identity of purpose," Bauer & Cie v. O'Donnell, 229 U.S. 1, 13, 33 S.Ct. 616, 57 L.Ed. 1041 (1913), and many everyday products-"automobiles, microwaves, calculators, mobile phones, tablets, and personal computers"-are subject to both patent and copyright protections, see Kirtsaeng, 568 U.S., at 545, 133 S.Ct. 1351 ; Brief for Costco Wholesale Corp. et al. as Amici Curiae 14-15. There is a "historic kinship between patent law and copyright law," Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417, 439, 104 S.Ct. 774, 78 L.Ed.2d 574 (1984), and the bond between the two leaves no room for a rift on the question of international exhaustion.

Lexmark sees the matter differently. The Patent Act, it points out, limits the patentee's "right to exclude others" from making, using, selling, or importing its products to acts that occur in the United States. 35 U.S.C. § 154(a). A domestic sale, it argues, triggers exhaustion because the sale compensates the patentee for "surrendering [those] U.S. rights." Brief for Respondent 38. A foreign sale is different: The Patent Act does not give patentees exclusionary powers abroad. Without those powers, a patentee selling in a foreign market may not be able to sell its product for the same price that it could in the United States, and therefore is not sure to receive "the reward guaranteed by U.S. patent law." Id., at 39 (internal quotation marks omitted). Absent that reward, says Lexmark, there should be no exhaustion. In short, there is no patent exhaustion from sales abroad because there are no patent rights abroad to exhaust.

The territorial limit on patent rights is, however, no basis for distinguishing copyright protections; those protections *1537"do not have any extraterritorial operation" either. 5 M. Nimmer & D. Nimmer, Copyright § 17.02, p. 17-26 (2017). Nor does the territorial limit support the premise of Lexmark's argument. Exhaustion is a separate limit on the patent grant, and does not depend on the patentee receiving some undefined premium for selling the right to access the American market. A purchaser buys an item, not patent rights. And exhaustion is triggered by the patentee's decision to give that item up and receive whatever fee it decides is appropriate "for the article and the invention which it embodies." Univis, 316 U.S., at 251, 62 S.Ct. 1088. The patentee may not be able to command the same amount for its products abroad as it does in the United States. But the Patent Act does not guarantee a particular price, much less the price from selling to American consumers. Instead, the right to exclude just ensures that the patentee receives one reward-of whatever amount the patentee deems to be "satisfactory compensation," Keeler, 157 U.S., at 661, 15 S.Ct. 738 -for every item that passes outside the scope of the patent monopoly.

This Court has addressed international patent exhaustion in only one case, Boesch v. Graff, decided over 125 years ago. All that case illustrates is that a sale abroad does not exhaust a patentee's rights when the patentee had nothing to do with the transaction. Boesch -from the days before the widespread adoption of electrical lighting-involved a retailer who purchased lamp burners from a manufacturer in Germany, with plans to sell them in the United States. The manufacturer had authority to make the burners under German law, but there was a hitch: Two individuals with no ties to the German manufacturer held the American patent to that invention. These patentees sued the retailer for infringement when the retailer imported the lamp burners into the United States, and we rejected the argument that the German manufacturer's sale had exhausted the American patentees' rights. The German manufacturer had no permission to sell in the United States from the American patentees, and the American patentees had not exhausted their patent rights in the products because they had not sold them to anyone, so "purchasers from [the German manufacturer] could not be thereby authorized to sell the articles in the United States." 133 U.S. 697, 703, 10 S.Ct. 378, 33 L.Ed. 787 (1890).

Our decision did not, as Lexmark contends, exempt all foreign sales from patent exhaustion. See Brief for Respondent 44-45. Rather, it reaffirmed the basic premise that only the patentee can decide whether to make a sale that exhausts its patent rights in an item. The American patentees did not do so with respect to the German products, so the German sales did not exhaust their rights.

Finally, the United States, as an amicus, advocates what it views as a middle-ground position: that "a foreign sale authorized by the U.S. patentee exhausts U.S. patent rights unless those rights are expressly reserved." Brief for United States 7-8. Its position is largely based on policy rather than principle. The Government thinks that an overseas "buyer's legitimate expectation" is that a "sale conveys all of the seller's interest in the patented article," so the presumption should be that a foreign sale triggers exhaustion. Id., at 32-33. But, at the same time, "lower courts long ago coalesced around" the rule that "a patentee's express reservation of U.S. patent rights at the time of a foreign sale will be given effect," so that option should remain open to the patentee. Id., at 22 (emphasis deleted).

The Government has little more than "long ago" on its side. In the 1890s, two *1538circuit courts-in cases involving the same company-did hold that patentees may use express restrictions to reserve their patent rights in connection with foreign sales. See Dickerson v. Tinling, 84 F. 192, 194-195 (C.A.8 1897) ; Dickerson v. Matheson, 57 F. 524, 527 (C.A.2 1893). But no "coalesc[ing]" ever took place: Over the following hundred-plus years, only a smattering of lower court decisions mentioned this express-reservation rule for foreign sales. See, e.g ., Sanofi, S.A. v. Med-Tech Veterinarian Prods., Inc., 565 F.Supp. 931, 938 (D.N.J.1983). And in 2001, the Federal Circuit adopted its blanket rule that foreign sales do not trigger exhaustion, even if the patentee fails to expressly reserve its rights. Jazz Photo, 264 F.3d, at 1105. These sparse and inconsistent decisions provide no basis for any expectation, let alone a settled one, that patentees can reserve patent rights when they sell abroad.

The theory behind the Government's express-reservation rule also wrongly focuses on the likely expectations of the patentee and purchaser during a sale. Exhaustion does not arise because of the parties' expectations about how sales transfer patent rights. More is at stake when it comes to patents than simply the dealings between the parties, which can be addressed through contract law. Instead, exhaustion occurs because, in a sale, the patentee elects to give up title to an item in exchange for payment. Allowing patent rights to stick remora-like to that item as it flows through the market would violate the principle against restraints on alienation. Exhaustion does not depend on whether the patentee receives a premium for selling in the United States, or the type of rights that buyers expect to receive. As a result, restrictions and location are irrelevant; what matters is the patentee's decision to make a sale.

* * *

The judgment of the United States Court of Appeals for the Federal Circuit is reversed, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Justice GORSUCH took no part in the consideration or decision of this case.

Justice GINSBURG, concurring in part and dissenting in part.

I concur in the Court's holding regarding domestic exhaustion-a patentee who sells a product with an express restriction on reuse or resale may not enforce that restriction through an infringement lawsuit, because the U.S. sale exhausts the U.S. patent rights in the product sold. See ante, at 1531 - 1536. I dissent, however, from the Court's holding on international exhaustion. A foreign sale, I would hold, does not exhaust a U.S. inventor's U.S. patent rights.

Patent law is territorial. When an inventor receives a U.S. patent, that patent provides no protection abroad. See Deepsouth Packing Co. v. Laitram Corp., 406 U.S. 518, 531, 92 S.Ct. 1700, 32 L.Ed.2d 273 (1972) ("Our patent system makes no claim to extraterritorial effect."). See also 35 U.S.C. § 271(a) (establishing liability for acts of patent infringement "within the United States" and for "import[ation] into the United States [of] any patented invention"). A U.S. patentee must apply to each country in which she seeks the exclusive right to sell her invention. Microsoft Corp. v. AT & T Corp., 550 U.S. 437, 456, 127 S.Ct. 1746, 167 L.Ed.2d 737 (2007) ("[F]oreign law alone, not United States law, currently governs the manufacture and sale of components of patented inventions in foreign countries."). See also Convention at Brussels, An Additional Act *1539Modifying the Paris Convention for the Protection of Industrial Property of Mar. 20, 1883, Dec. 14, 1900, Art. I, 32 Stat. 1940 ("Patents applied for in the different contracting States ... shall be independent of the patents obtained for the same invention in the other States."). And patent laws vary by country; each country's laws "may embody different policy judgments about the relative rights of inventors, competitors, and the public in patented inventions." Microsoft, 550 U.S., at 455, 127 S.Ct. 1746 (internal quotation marks omitted).

Because a sale abroad operates independently of the U.S. patent system, it makes little sense to say that such a sale exhausts an inventor's U.S. patent rights. U.S. patent protection accompanies none of a U.S. patentee's sales abroad-a competitor could sell the same patented product abroad with no U.S.-patent-law consequence. Accordingly, the foreign sale should not diminish the protections of U.S. law in the United States.

The majority disagrees, in part because this Court decided, in Kirtsaeng v. John Wiley & Sons, Inc., 568 U.S. 519, 525, 133 S.Ct. 1351, 185 L.Ed.2d 392 (2013), that a foreign sale exhausts U.S. copyright protections. Copyright and patent exhaustion, the majority states, "share a strong similarity." Ante, at 1536 (internal quotation marks omitted). I dissented from our decision in Kirtsaeng and adhere to the view that a foreign sale should not exhaust U.S. copyright protections. See 568 U.S., at 557, 133 S.Ct. 1351.

But even if I subscribed to Kirtsaeng 's reasoning with respect to copyright, that decision should bear little weight in the patent context. Although there may be a "historical kinship" between patent law and copyright law, Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417, 439, 104 S.Ct. 774, 78 L.Ed.2d 574 (1984), the two "are not identical twins," id., at 439, n. 19, 104 S.Ct. 774. The Patent Act contains no analogue to 17 U.S.C. § 109(a), the Copyright Act first-sale provision analyzed in Kirtsaeng . See ante, at 1535 - 1536. More importantly, copyright protections, unlike patent protections, are harmonized across countries. Under the Berne Convention, which 174 countries have joined,* members "agree to treat authors from other member countries as well as they treat their own." Golan v. Holder, 565 U.S. 302, 308, 132 S.Ct. 873, 181 L.Ed.2d 835 (2012) (citing Berne Convention for the Protection of Literary and Artistic Works, Sept. 9, 1886, as revised at Stockholm on July 14, 1967, Arts. 1, 5(1), 828 U.N.T.S. 225, 231-233). The copyright protections one receives abroad are thus likely to be similar to those received at home, even if provided under each country's separate copyright regime.

For these reasons, I would affirm the Federal Circuit's judgment with respect to foreign exhaustion.