Justice Ginsburg delivered the opinion of the Court.

[1] This copyright case concerns the rights of freelance authors and a presumptive privilege of their publishers. The litigation was initiated by six freelance authors and relates to articles they contributed to three print periodicals (two newspapers and one magazine). Under agreements with the periodicals’ publishers, but without the freelancers’ consent, two computer database companies placed copies of the freelancers’ articles–along with all other articles from the periodicals in which the freelancers’ work appeared–into three databases. Whether written by a freelancer or staff member, each article is presented to, and retrievable by, the user in isolation, clear of the context the original print publication presented.

[2] The freelance authors’ complaint alleged that their copyrights had been infringed by the inclusion of their articles in the databases. The publishers, in response, relied on the privilege of reproduction and distribution accorded them by §201(c) of the Copyright Act, which provides:

“Copyright in each separate contribution to a collective work is distinct from copyright in the collective work as a whole, and vests initially in the author of the contribution. In the absence of an express transfer of the copyright or of any rights under it, the owner of copyright in the collective work is presumed to have acquired only the privilege of reproducing and distributing the contribution as part of that particular collective work, any revision of that collective work, and any later collective work in the same series.”

17 U.S.C. § 201(c).

[3] Specifically, the publishers maintained that, as copyright owners of collective works, i.e., the original print publications, they had merely exercised “the privilege” §201(c) accords them to “reproduc[e] and distribut[e]” the author’s discretely copyrighted contribution.

[4] In agreement with the Second Circuit, we hold that §201(c) does not authorize the copying at issue here. The publishers are not sheltered by §201(c), we conclude, because the databases reproduce and distribute articles standing alone and not in context, not “as part of that particular collective work” to which the author contributed, “as part of … any revision” thereof, or “as part of … any later collective work in the same series.” Both the print publishers and the electronic publishers, we rule, have infringed the copyrights of the freelance authors.

I

A

[5] Respondents Jonathan Tasini, Mary Kay Blakely, Barbara Garson, Margot Mifflin, Sonia Jaffe Robbins, and David S. Whitford are authors (Authors). Between 1990 and 1993, they wrote the 21 articles (Articles) on which this dispute centers. Tasini, Mifflin, and Blakely contributed 12 Articles to The New York Times, the daily newspaper published by petitioner The New York Times Company (Times). Tasini, Garson, Robbins, and Whitford wrote eight Articles for Newsday, another New York daily paper, published by petitioner Newsday, Inc. (Newsday). Whitford also contributed one Article to Sports Illustrated, a weekly magazine published by petitioner Time, Inc. (Time). The Authors registered copyrights in each of the Articles. The Times, Newsday, and Time (Print Publishers) registered collective work copyrights in each periodical edition in which an Article originally appeared. The Print Publishers engaged the Authors as independent contractors (freelancers) under contracts that in no instance secured consent from an Author to placement of an Article in an electronic database. [n1]

[6] At the time the Articles were published, all three Print Publishers had agreements with petitioner LEXIS/NEXIS (formerly Mead Data Central Corp.), owner and operator of NEXIS, a computerized database that stores information in a text-only format. NEXIS contains articles from hundreds of journals (newspapers and periodicals) spanning many years. The Print Publishers have licensed to LEXIS/NEXIS the text of articles appearing in the three periodicals. The licenses authorize LEXIS/NEXIS to copy and sell any portion of those texts.

[7] Pursuant to the licensing agreements, the Print Publishers regularly provide LEXIS/NEXIS with a batch of all the articles published in each periodical edition. The Print Publisher codes each article to facilitate computerized retrieval, then transmits it in a separate file. After further coding, LEXIS/NEXIS places the article in the central discs of its database.

[8] Subscribers to NEXIS, accessing the system through a computer, may search for articles by author, subject, date, publication, headline, key term, words in text, or other criteria. Responding to a search command, NEXIS scans the database and informs the user of the number of articles meeting the user’s search criteria. The user then may view, print, or download each of the articles yielded by the search. The display of each article includes the print publication (e.g., The New York Times), date (September 23, 1990), section (Magazine), initial page number (26), headline or title (“Remembering Jane”), and author (Mary Kay Blakely). Each article appears as a separate, isolated “story”–without any visible link to the other stories originally published in the same newspaper or magazine edition. NEXIS does not contain pictures or advertisements, and it does not reproduce the original print publication’s formatting features such as headline size, page placement (e.g., above or below the fold for newspapers), or location of continuation pages.

[9] The Times (but not Newsday or Time) also has licensing agreements with petitioner University Microfilms International (UMI). The agreements authorize reproduction of Times materials on two CD-ROM products, the New York Times OnDisc (NYTO) and General Periodicals OnDisc (GPO).

[10] Like NEXIS, NYTO is a text-only system. Unlike NEXIS, NYTO, as its name suggests, contains only the Times. Pursuant to a three-way agreement, LEXIS/NEXIS provides UMI with computer files containing each article as transmitted by the Times to LEXIS/NEXIS. Like LEXIS/NEXIS, UMI marks each article with special codes. UMI also provides an index of all the articles in NYTO. Articles appear in NYTO in essentially the same way they appear in NEXIS, i.e., with identifying information (author, title, etc.), but without original formatting or accompanying images.

[11] GPO contains articles from approximately 200 publications or sections of publications. Unlike NEXIS and NYTO, GPO is an image-based, rather than a text-based, system. The Times has licensed GPO to provide a facsimile of the Times’ Sunday Book Review and Magazine. UMI “burns” images of each page of these sections onto CD-ROMs. The CD-ROMs show each article exactly as it appeared on printed pages, complete with photographs, captions, advertisements, and other surrounding materials. UMI provides an index and abstracts of all the articles in GPO.

[12] Articles are accessed through NYTO and GPO much as they are accessed through NEXIS. The user enters a search query using similar criteria (e.g., author, headline, date). The computer program searches available indexes and abstracts, and retrieves a list of results matching the query. The user then may view each article within the search result, and may print the article or download it to a disc. The display of each article provides no links to articles appearing on other pages of the original print publications. [n2]

B

[13] On December 16, 1993, the Authors filed this civil action in the United States District Court for the Southern District of New York. The Authors alleged that their copyrights were infringed when, as permitted and facilitated by the Print Publishers, LEXIS/NEXIS and UMI (Electronic Publishers) placed the Articles in the NEXIS, NYTO, and GPO databases (Databases). The Authors sought declaratory and injunctive relief, and damages. In response to the Authors’ complaint, the Print and Electronic Publishers raised the reproduction and distribution privilege accorded collective work copyright owners by 17 U.S.C. § 201(c). After discovery, both sides moved for summary judgment.

[14] The District Court granted summary judgment for the Publishers, holding that §201(c) shielded the Database reproductions. 972 F. Supp. 804, 806 (1997). The privilege conferred by §201(c) is transferable, the court first concluded, and therefore could be conveyed from the original Print Publishers to the Electronic Publishers. Id., at 816. Next, the court determined, the Databases reproduced and distributed the Authors’ works, in §201(c)’s words, “as part of … [a] revision of that collective work” to which the Authors had first contributed. To qualify as “revisions,” according to the court, works need only “preserve some significant original aspect of [collective works]–whether an original selection or an original arrangement.” Id., at 821. This criterion was met, in the District Court’s view, because the Databases preserved the Print Publishers’ “selection of articles” by copying all of the articles originally assembled in the periodicals’ daily or weekly issues. Id., at 823. The Databases “highlight[ed]” the connection between the articles and the print periodicals, the court observed, by showing for each article not only the author and periodical, but also the print publication’s particular issue and page numbers. Id., at 824 (“[T]he electronic technologies not only copy the publisher defendants’ complete original ‘selection’ of articles, they tag those articles in such a way that the publisher defendants’ original selection remains evident online.”).

[15] The Authors appealed, and the Second Circuit reversed. 206 F.3d 161 (1999). The Court of Appeals granted summary judgment for the Authors on the ground that the Databases were not among the collective works covered by §201(c), and specifically, were not “revisions” of the periodicals in which the Articles first appeared. Id., at 167-170. Just as §201(c) does not “permit a Publisher to sell a hard copy of an Author’s article directly to the public even if the Publisher also offered for individual sale all of the other articles from the particular edition,” the court reasoned, so §201(c) does not allow a Publisher to “achieve the same goal indirectly” through computer databases. Id., at 168. In the Second Circuit’s view, the Databases effectively achieved this result by providing multitudes of “individually retrievable” articles. Ibid. As stated by the Court of Appeals, the Databases might fairly be described as containing “new antholog[ies] of innumerable” editions or publications, but they do not qualify as “revisions” of particular editions of periodicals in the Databases. Id., at 169. Having concluded that §201(c) “does not permit the Publishers,” acting without the author’s consent, “to license individually copyrighted works for inclusion in the electronic databases,” the court did not reach the question whether the §201(c) privilege is transferable. Id., at 165, and n. 2.

[16] We granted certiorari to determine whether the copying of the Authors’ Articles in the Databases is privileged by 17 U.S.C. § 201(c). 531 U.S. 978 (2000). Like the Court of Appeals, we conclude that the §201(c) privilege does not override the Authors’ copyrights, for the Databases do not reproduce and distribute the Articles as part of a collective work privileged by §201(c). Accordingly, and again like the Court of Appeals, we find it unnecessary to determine whether the privilege is transferable.

[17] Under the Copyright Act, as amended in 1976, “[c]opyright protection subsists … in original works of authorship fixed in any tangible medium of expression … from which they can be perceived, reproduced, or otherwise communicated.” 17 U.S.C. § 102(a). When, as in this case, a freelance author has contributed an article to a “collective work” such as a newspaper or magazine, see §101 (defining “collective work”), the statute recognizes two distinct copyrighted works: “Copyright in each separate contribution to a collective work is distinct from copyright in the collective work as a whole … .” §201(c) (emphasis added). Copyright in the separate contribution “vests initially in the author of the contribution” (here, the freelancer). Ibid. Copyright in the collective work vests in the collective author (here, the newspaper or magazine publisher) and extends only to the creative material contributed by that author, not to “the preexisting material employed in the work,” §103(b). See also Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340, 358 (1991) (copyright in “compilation”–a term that includes “collective works,” 17 U.S.C. § 101–is limited to the compiler’s original “selection, coordination, and arrangement”).

[18] Prior to the 1976 revision, as the courts below recognized, see 206 F.3d, at 168; 972 F. Supp., at 815, authors risked losing their rights when they placed an article in a collective work. Pre-1976 copyright law recognized a freelance author’s copyright in a published article only when the article was printed with a copyright notice in the author’s name. See Copyright Act of 1909, §18, 35 Stat. 1079. When publishers, exercising their superior bargaining power over authors, declined to print notices in each contributor’s name, the author’s copyright was put in jeopardy. See A. Kaminstein, Divisibility of Copyrights, Study No. 11, in Copyright Law Revision Studies Nos. 11-13, prepared for the Senate Committee on the Judiciary, 86th Cong., 2d Sess., p. 18 (1960). The author did not have the option to assign only the right of publication in the periodical; such a partial assignment was blocked by the doctrine of copyright “indivisibility.” See id., at 11. Thus, when a copyright notice appeared only in the publisher’s name, the author’s work would fall into the public domain, unless the author’s copyright, in its entirety, had passed to the publisher. See id., at 18. Such complete transfer might be accomplished by a contract, perhaps one with a provision, not easily enforced, for later retransfer of rights back to the author. See id., at 20-22. Or, absent a specific contract, a court might find that an author had tacitly transferred the entire copyright to a publisher, in turn deemed to hold the copyright in “trust” for the author’s benefit. See id., at 18-19; see generally 3 M. Nimmer, Copyright §10.01[C][2], pp. 10-12 to 10-14 (2000).

[19] In the 1976 revision, Congress acted to “clarify and improve [this] confused and frequently unfair legal situation with respect to rights in contributions.” H. R. Rep. No. 94-1476, p. 122 (1976) (hereinafter H. R. Rep.). [n3] The 1976 Act rejected the doctrine of indivisibility, recasting the copyright as a bundle of discrete “exclusive rights,” 17 U.S.C. § 106 (1994 ed. and Supp. V), [n4] each of which “may be transferred … and owned separately,” §201(d)(2). [n5] Congress also provided, in §404(a), that “a single notice applicable to the collective work as a whole is sufficient” to protect the rights of freelance contributors. And in §201(c), Congress codified the discrete domains of “[c]opyright in each separate contribution to a collective work” and “copyright in the collective work as a whole.” Together, §404(a) and §201(c) “preserve the author’s copyright in a contribution even if the contribution does not bear a separate notice in the author’s name, and without requiring any unqualified transfer of rights to the owner of the collective work.” H. R. Rep. 122.

[20] Section 201(c) both describes and circumscribes the “privilege” a publisher acquires regarding an author’s contribution to a collective work:

“In the absence of an express transfer of the copyright or of any rights under it, the owner of copyright in the collective work is presumed to have acquired only the privilege of reproducing and distributing the contribution as part of that particular collective work, any revision of that collective work, and any later collective work in the same series.” (Emphasis added.)

A newspaper or magazine publisher is thus privileged to reproduce or distribute an article contributed by a freelance author, absent a contract otherwise providing, only “as part of” any (or all) of three categories of collective works: (a) “that collective work” to which the author contributed her work, (b) “any revision of that collective work,” or (c) “any later collective work in the same series.” In accord with Congress’ prescription, a “publishing company could reprint a contribution from one issue in a later issue of its magazine, and could reprint an article from a 1980 edition of an encyclopedia in a 1990 revision of it; the publisher could not revise the contribution itself or include it in a new anthology or an entirely different magazine or other collective work.” H. R. Rep. 122-123.

[21] Essentially, §201(c) adjusts a publisher’s copyright in its collective work to accommodate a freelancer’s copyright in her contribution. If there is demand for a freelance article standing alone or in a new collection, the Copyright Act allows the freelancer to benefit from that demand; after authorizing initial publication, the freelancer may also sell the article to others. Cf. Stewart v. Abend, 495 U.S. 207, 229 (1990) (“[w]hen an author produces a work which later commands a higher price in the market than the original bargain provided, the copyright statute [i.e., the separate renewal term of former 17 U.S.C. § 24] is designed to provide the author the power to negotiate for the realized value of the work”); id., at 230 (noting author’s “inalienable termination right” under current 17 U.S.C. § 203, 302). It would scarcely “preserve the author’s copyright in a contribution” as contemplated by Congress, H. R. Rep. 122, if a newspaper or magazine publisher were permitted to reproduce or distribute copies of the author’s contribution in isolation or within new collective works. See Gordon, Fine-Tuning Tasini: Privileges of Electronic Distribution and Reproduction, 66 Brooklyn L. Rev. 473, 484 (2000). [n6]

III

[22] In the instant case, the Authors wrote several Articles and gave the Print Publishers permission to publish the Articles in certain newspapers and magazines. It is undisputed that the Authors hold copyrights and, therefore, exclusive rights in the Articles. [n7] It is clear, moreover, that the Print and Electronic Publishers have exercised at least some rights that §106 initially assigns exclusively to the Authors: LEXIS/NEXIS’ central discs and UMI’s CD-ROMs “reproduce … copies” of the Articles, §106(1); UMI, by selling those CD-ROMs, and LEXIS/NEXIS, by selling copies of the Articles through the NEXIS Database, “distribute copies” of the Articles “to the public by sale,” §106(3); and the Print Publishers, through contracts licensing the production of copies in the Databases, “authorize” reproduction and distribution of the Articles, §106. [n8]

[23] Against the Authors’ charge of infringement, the Publishers do not here contend the Authors entered into an agreement authorizing reproduction of the Articles in the Databases. See supra, at 3, n. 1. Nor do they assert that the copies in the Databases represent “fair use” of the Authors’ Articles. See 17 U.S.C. § 107 (“fair use of a copyrighted work … is not an infringement”; four factors identified among those relevant to fair use determination). Instead, the Publishers rest entirely on the privilege described in §201(c). Each discrete edition of the periodicals in which the Articles appeared is a “collective work,” the Publishers agree. They contend, however, that reproduction and distribution of each Article by the Databases lie within the “privilege of reproducing and distributing the [Articles] as part of … [a] revision of that collective work,” §201(c). The Publishers’ encompassing construction of the §201(c) privilege is unacceptable, we conclude, for it would diminish the Authors’ exclusive rights in the Articles.

[24] In determining whether the Articles have been reproduced and distributed “as part of” a “revision” of the collective works in issue, we focus on the Articles as presented to, and perceptible by, the user of the Databases. See §102 (copyright protection subsists in original works fixed in any medium “from which they can be perceived, reproduced, or otherwise communicated”); see also §101 (definitions of “copies” and “fixed”); Haemmerli, Commentary: Tasini v. New York Times Co., 22 Colum.-VLA. J. L. & Arts 129, 142-143 (1998). In this case, the three Databases present articles to users clear of the context provided either by the original periodical editions or by any revision of those editions. The Databases first prompt users to search the universe of their contents: thousands or millions of files containing individual articles from thousands of collective works (i.e., editions), either in one series (the Times, in NYTO) or in scores of series (the sundry titles in NEXIS and GPO). When the user conducts a search, each article appears as a separate item within the search result. In NEXIS and NYTO, an article appears to a user without the graphics, formatting, or other articles with which the article was initially published. In GPO, the article appears with the other materials published on the same page or pages, but without any material published on other pages of the original periodical. In either circumstance, we cannot see how the Database perceptibly reproduces and distributes the article “as part of ” either the original edition or a “revision” of that edition.

[25] One might view the articles as parts of a new compendium–namely, the entirety of works in the Database. In that compendium, each edition of each periodical represents only a miniscule fraction of the ever-expanding Database. The Database no more constitutes a “revision” of each constituent edition than a 400-page novel quoting a sonnet in passing would represent a “revision” of that poem. “Revision” denotes a new “version,” and a version is, in this setting, a “distinct form of something regarded by its creators or others as one work.” Webster’s Third New International Dictionary 1944, 2545 (1976). The massive whole of the Database is not recognizable as a new version of its every small part.

[26] Alternatively, one could view the Articles in the Databases “as part of” no larger work at all, but simply as individual articles presented individually. That each article bears marks of its origin in a particular periodical (less vivid marks in NEXIS and NYTO, more vivid marks in GPO) suggests the article was previously part of that periodical. But the markings do not mean the article is currently reproduced or distributed as part of the periodical. The Databases’ reproduction and distribution of individual Articles–simply as individual Articles–would invade the core of the Authors’ exclusive rights under §106. [n9]

[27] The Publishers press an analogy between the Databases, on the one hand, and microfilm and microfiche, on the other. We find the analogy wanting. Microforms typically contain continuous photographic reproductions of a periodical in the medium of miniaturized film. Accordingly, articles appear on the microforms, writ very small, in precisely the position in which the articles appeared in the newspaper. The Times, for example, printed the beginning of Blakely’s “Remembering Jane” Article on page 26 of the Magazine in the September 23, 1990, edition; the microfilm version of the Times reproduces that same Article on film in the very same position, within a film reproduction of the entire Magazine, in turn within a reproduction of the entire September 23, 1990, edition. True, the microfilm roll contains multiple editions, and the microfilm user can adjust the machine lens to focus only on the Article, to the exclusion of surrounding material. Nonetheless, the user first encounters the Article in context. In the Databases, by contrast, the Articles appear disconnected from their original context. In NEXIS and NYTO, the user sees the “Jane” Article apart even from the remainder of page 26. In GPO, the user sees the Article within the context of page 26, but clear of the context of page 25 or page 27, the rest of the Magazine, or the remainder of the day’s newspaper. In short, unlike microforms, the Databases do not perceptibly reproduce articles as part of the collective work to which the author contributed or as part of any “revision” thereof. [n10]

[28] Invoking the concept of “media neutrality,” the Publishers urge that the “transfer of a work between media” does not “alte[r] the character of” that work for copyright purposes. Brief for Petitioners 23. That is indeed true. See 17 U.S.C. § 102(a) (copyright protection subsists in original works “fixed in any tangible medium of expression”). But unlike the conversion of newsprint to microfilm, the transfer of articles to the Databases does not represent a mere conversion of intact periodicals (or revisions of periodicals) from one medium to another. The Databases offer users individual articles, not intact periodicals. In this case, media neutrality should protect the Authors’ rights in the individual Articles to the extent those Articles are now presented individually, outside the collective work context, within the Databases’ new media. [n11]

[29] For the purpose at hand–determining whether the Authors’ copyrights have been infringed–an analogy to an imaginary library may be instructive. [n12] Rather than maintaining intact editions of periodicals, the library would contain separate copies of each article. Perhaps these copies would exactly reproduce the periodical pages from which the articles derive (if the model is GPO); perhaps the copies would contain only typescript characters, but still indicate the original periodical’s name and date, as well as the article’s headline and page number (if the model is NEXIS or NYTO). The library would store the folders containing the articles in a file room, indexed based on diverse criteria, and containing articles from vast numbers of editions. In response to patron requests, an inhumanly speedy librarian would search the room and provide copies of the articles matching patron-specified criteria.

[30] Viewing this strange library, one could not, consistent with ordinary English usage, characterize the articles “as part of” a “revision” of the editions in which the articles first appeared. In substance, however, the Databases differ from the file room only to the extent they aggregate articles in electronic packages (the LEXIS/NEXIS central discs or UMI CD-ROMs), while the file room stores articles in spatially separate files. The crucial fact is that the Databases, like the hypothetical library, store and retrieve articles separately within a vast domain of diverse texts. Such a storage and retrieval system effectively overrides the Authors’ exclusive right to control the individual reproduction and distribution of each Article, 17 U.S.C. § 106(1), (3). Cf. Ryan v. Carl Corp., 23 F. Supp. 2d 1146 (ND Cal. 1998) (holding copy shop in violation of §201(c)).

[31] The Publishers claim the protection of §201(c) because users can manipulate the Databases to generate search results consisting entirely of articles from a particular periodical edition. By this logic, §201(c) would cover the hypothetical library if, in response to a request, that library’s expert staff assembled all of the articles from a particular periodical edition. However, the fact that a third party can manipulate a database to produce a noninfringing document does not mean the database is not infringing. Under §201(c), the question is not whether a user can generate a revision of a collective work from a database, but whether the database itself perceptibly presents the author’s contribution as part of a revision of the collective work. That result is not accomplished by these Databases.

[32] The Publishers finally invoke Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417 (1984). That decision, however, does not genuinely aid their argument. Sony held that the “sale of copying equipment” does not constitute contributory infringement if the equipment is “capable of substantial noninfringing uses.” Id., at 442. The Publishers suggest that their Databases could be liable only under a theory of contributory infringement, based on end-user conduct, which the Authors did not plead. The Electronic Publishers, however, are not merely selling “equipment”; they are selling copies of the Articles. And, as we have explained, it is the copies themselves, without any manipulation by users, that fall outside the scope of the §201(c) privilege.

IV

[33] The Publishers warn that a ruling for the Authors will have “devastating” consequences. Brief for Petitioners 49. The Databases, the Publishers note, provide easy access to complete newspaper texts going back decades. A ruling for the Authors, the Publishers suggest, will punch gaping holes in the electronic record of history. The Publishers’ concerns are echoed by several historians, see Brief for Ken Burns et al. as Amici Curiae, but discounted by several other historians, see Brief for Ellen Schrecker et al. as Amici Curiae; Brief for Authors’ Guild, Jacques Barzun et al. as Amici Curiae.

[34] Notwithstanding the dire predictions from some quarters, see also post, at 16 (Stevens, J., dissenting), it hardly follows from today’s decision that an injunction against the inclusion of these Articles in the Databases (much less all freelance articles in any databases) must issue. See 17 U.S.C. § 502(a) (court “may” enjoin infringement); Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 578, n. 10 (1994) (goals of copyright law are “not always best served by automatically granting injunctive relief”). The parties (Authors and Publishers) may enter into an agreement allowing continued electronic reproduction of the Authors’ works; they, and if necessary the courts and Congress, may draw on numerous models for distributing copyrighted works and remunerating authors for their distribution. See, e.g., 17 U.S.C. § 118(b); Broadcast Music, Inc. v. Columbia Broadcasting System, Inc., 441 U.S. 1, 4-6, 10-12 (1979) (recounting history of blanket music licensing regimes and consent decrees governing their operation). [n13] In any event, speculation about future harms is no basis for this Court to shrink authorial rights Congress established in §201(c). Agreeing with the Court of Appeals that the Publishers are liable for infringement, we leave remedial issues open for initial airing and decision in the District Court.

[35] We conclude that the Electronic Publishers infringed the Authors’ copyrights by reproducing and distributing the Articles in a manner not authorized by the Authors and not privileged by §201(c). We further conclude that the Print Publishers infringed the Authors’ copyrights by authorizing the Electronic Publishers to place the Articles in the Databases and by aiding the Electronic Publishers in that endeavor. We therefore affirm the judgment of the Court of Appeals.

[36] It is so ordered.

Justice Stevens, with whom Justice Breyer joins, dissenting.

[37] This case raises an issue of first impression concerning the meaning of the word “revision” as used in §201(c) of the 1976 revision of the Copyright Act of 1909 (1976 Act). Ironically, the Court today seems unwilling to acknowledge that changes in a collective work far less extensive than those made to prior copyright law by the 1976 “revision” do not merit the same characterization.

[38] To explain my disagreement with the Court’s holding, I shall first identify Congress’ principal goals in passing the 1976 Act’s changes in the prior law with respect to collective works. I will then discuss two analytically separate questions that are blended together in the Court’s discussion of revisions. The first is whether the electronic versions of the collective works created by the owners of the copyright in those works (Print Publishers or publishers) are “revision[s]” of those works within the meaning of 17 U.S.C. § 201(c). In my judgment they definitely are. The second is whether the aggregation by LEXIS/NEXIS and UMI (Electronic Databases) of the revisions with other editions of the same periodical or with other periodicals within a single database changes the equation. I think it does not. Finally, I will consider the implications of broader copyright policy for the issues presented in this case.

I

[39] As the majority correctly observes, prior to 1976, an author’s decision to publish her individual article as part of a collective work was a perilous one. Although pre-1976 copyright law recognized the author’s copyright in an individual article that was included within a collective work, those rights could be lost if the publisher refused to print the article with a copyright notice in the author’s name. 3 M. Nimmer & D. Nimmer, Nimmer on Copyright §10.01[C][2], p. 10-12 (2001).

[40] This harsh rule was, from the author’s point of view, exacerbated by the pre-1976 doctrine of copyright “indivisibility,” which prevented an author from assigning only limited publication rights to the publisher of a collective work while holding back all other rights to herself. [n1] Ibid. The indivisibility of copyright, in combination with the danger of losing copyright protection, put significant pressure on an author seeking to preserve her copyright in the contribution to transfer the entire copyright over to the publisher in trust. See Kaminstein, Divisibility of Copyrights, Study No. 11, in Copyright Law Revision Studies Nos. 11-13, prepared for the Senate Committee on the Judiciary, 86th Cong., 2d Sess., 18-22 (1960) (herinafter Kaminstein). [n2] Such authors were often at the mercy of publishers when they tried to reclaim their copyright. Id., at 21. [n3]

[41] The 1976 Act’s extensive revisions of the copyright law had two principal goals with respect to the rights of freelance authors whose writings appeared as part of larger collective works. First, as the legislative history of §201(c) unambiguously reveals, one of its most significant aims was to “preserve the author’s copyright in a contribution even if the contribution does not bear a separate notice in the author’s name, and without requiring any unqualified transfer of rights to the owner of the collective work.” H. R. Rep. No. 94-1476, p. 122 (1976) (hereinafter H. R. Rep.) (discussing the purpose of §201(c)). Indeed, §404(a) states that “a single notice applicable to the collective work as a whole is sufficient” to protect the author’s rights.

[42] The second significant change effected by the 1976 Act clarified the scope of the privilege granted to the publisher of a collective work. While pre-1976 law had the effect of encouraging an author to transfer her entire copyright to the publisher of a collective work, §201(c) creates the opposite incentive, stating that, absent some agreement to the contrary, the publisher acquires from the author only “the privilege of reproducing and distributing the contribution as part of that particular collective work, any revision of that collective work, and any later collective work in the same series.” [n4] Congress intended this limitation on what the author is presumed to give away primarily to keep publishers from “revis[ing] the contribution itself or includ[ing] it in a new anthology or an entirely different magazine or other collective work.” H. R. Rep. 122-123. [n5]

[43] The majority is surely correct that the 1976 Act’s new approach to collective works was an attempt to “ ‘clarify and improve the … confused and frequently unfair legal situation’ ” that existed under the prior regime. Id., at 122. It is also undoubtedly true that the drafters of the 1976 Act hoped to “enhance the author’s position vis-à-vis the patron.” Ante, at 9, n. 3. It does not follow, however, that Congress’ efforts to “preserve the author’s copyright in a contribution,” H. R. Rep. 122, can only be honored by a finding in favor of the respondent authors.

[44] Indeed, the conclusion that the petitioners’ actions were lawful is fully consistent with both of Congress’ principal goals for collective works in the 1976 Act. First, neither the publication of the collective works by the Print Publishers, nor their transfer to the Electronic Databases had any impact on the legal status of the copyrights of the respondents’ individual contributions. [n6] By virtue of the 1976 Act, respondents remain the owners of the copyright in their individual works. Moreover, petitioners neither modified respondents’ individual contributions nor, as I will show in Part II, published them in a “new anthology or an entirely different magazine or other collective work.” H. R. Rep. 122-123 (emphasis added). Because I do not think it is at all obvious that the decision the majority reaches today is a result clearly intended by the 1976 Congress, I disagree with the Court’s conclusion that a ruling in petitioners’ favor would “shrink authorial rights” that “Congress [has] established.” Ante, at 21 (emphasis added).

II

[45] Not only is petitioners’ position consistent with Congress’ general goals in the 1976 Act, it is also consistent with the text of §201(c). That provision allows the publisher of a collective work to “reproduc[e] and distribut[e] the contribution as part of that particular collective work, any revision of that collective work, and any later collective work in the same series.” The central question in this case, then, is whether petitioners are correct when they argue that publication of the respondents’ articles in the various Electronic Databases at issue in this case is nothing more than “reproduc[tion] and distribut[ion] [of] the contribution as part of … revision[s] of [the original] collective work[s]” in which respondents’ articles appeared. I agree with petitioners that neither the conversion of the Print Publishers’ collective works from printed to electronic form, nor the transmission of those electronic versions of the collective works to the Electronic Databases, nor even the actions of the Electronic Databases once they receive those electronic versions does anything to deprive those electronic versions of their status as mere “revision[s]” of the original collective works.

[46] A proper analysis of this case benefits from an incremental approach. Accordingly, I begin by discussing an issue the majority largely ignores: whether a collection of articles from a single edition of the New York Times (i.e., the batch of files the Print Publishers periodically send to the Electronic Databases) constitutes a “revision” of an individual edition of the paper. In other words, does a single article within such a collection exist as “part of” a “revision”? Like the majority, I believe that the crucial inquiry is whether the article appears within the “context” of the original collective work. Ante, at 16. But this question simply raises the further issue of precisely how much “context” is enough.

[47] The record indicates that what is sent from the New York Times to the Electronic Databases (with the exception of General Periodicals on Disc (GPO)) is simply a collection of ASCII text files representing the editorial content of the New York Times for a particular day. [n7] App. 73a. Each individual ASCII file contains the text of a single article as well as additional coding intended to help readers identify the context in which the article originally appeared and to facilitate database searches. Thus, for example, to the original text of an article, the New York Times adds information on the article’s “headline, byline and title,” “the section of the paper in which the article had originally appeared,” and “the page in the paper or periodical on which the article had first appeared.” Id., at 75a-76a. [n8]

[48] I see no compelling reason why a collection of files corresponding to a single edition of the New York Times, standing alone, cannot constitute a “revision” of that day’s New York Times. It might be argued, as respondents appear to do, that the presentation of each article within its own electronic file makes it impossible to claim that the collection of files as a whole amounts to a “revision.” Brief for Respondents Tasini et al. 34. But the conversion of the text of the overall collective work into separate electronic files should not, by itself, decide the question. After all, one of the hallmarks of copyright policy, as the majority recognizes, ante, at 17, is the principle of media neutrality. See H. R. Rep. 53.

[49] No one doubts that the New York Times has the right to reprint its issues in Braille, in a foreign language, or in microform, even though such revisions might look and feel quite different from the original. Such differences, however, would largely result from the different medium being employed. Similarly, the decision to convert the single collective work newspaper into a collection of individual ASCII files can be explained as little more than a decision that reflects the different nature of the electronic medium. Just as the paper version of the New York Times is divided into “sections” and “pages” in order to facilitate the reader’s navigation and manipulation of large batches of newsprint, so too the decision to subdivide the electronic version of that collective work into individual article files facilitates the reader’s use of the electronic information. The bare-bones nature of ASCII text would make trying to wade through a single ASCII file containing the entire content of a single edition of the New York Times an exercise in frustration. [n9]

[50] Although the Court does not separately discuss the question whether the groups of files that the New York Times sends to the Electronic Databases constitute “revision[s],” its reasoning strongly suggests that it would not accept such a characterization. The majority, for example, places significant emphasis on the differences between the various Electronic Databases and microform, a medium that admittedly qualifies as a revision under §201(c). [n10] As with the conversion of individual editions into collections of separate article-files, however, many of the differences between the electronic versions and microform are necessitated by the electronic medium. The Court therefore appears to back away from principles of media neutrality when it implicitly criticizes ASCII-text files for their inability to reproduce “Remembering Jane” “in the very same position, within a film reproduction of the entire Magazine, in turn within a reproduction of the entire September 23, 1990, edition.” Ante, at 16. [n11]

[51] In contrast, I think that a proper respect for media neutrality suggests that the New York Times, reproduced as a collection of individual ASCII files, should be treated as a “revision” of the original edition, as long as each article explicitly refers to the original collective work and as long as substantially the rest of the collective work is, at the same time, readily accessible to the reader of the individual file. In this case, no one disputes that the first pieces of information a user sees when looking at an individual ASCII article file are the name of the publication in which the article appeared, the edition of that publication, and the location of the article within that edition. I agree with the majority that such labeling alone is insufficient to establish that the individual file exists as “part of” a revision of the original collective work. See ante, at 15. But such labeling is not all there is in the group of files sent to the Electronic Databases.

[52]  In addition to the labels, the batch of electronic files contains the entire editorial content of the original edition of the New York Times for that day. That is, while I might agree that a single article, standing alone, even when coded with identifying information (e.g., publication, edition date, headline, etc.), should not be characterized as a “part of” a larger collective work, I would not say the same about an individual article existing as “part of” a collection of articles containing all the editorial content of that day’s New York Times. This is all the more true because, as the District Court correctly noted, it is the Print Publishers’ selection process, the editorial process by which the staff of the New York Times, for example, decides which articles will be included in “All the News That’s Fit to Print,” that is the most important creative element they contribute to the collective works they publish. 972 F. Supp. 804, 823 (SDNY 1997). [n12] While such superficial features as page placement and column width are lost in ASCII format, the Print Publishers’ all-important editorial selection is wholly preserved in the collection of individual article-files sent to the Electronic Databases.

[53] To see why an electronic version of the New York Times made up of a group of individual ACSCII article-files, standing alone, may be considered a §201(c) revision, suppose that, instead of transmitting to NEXIS the articles making up a particular day’s edition, the New York Times saves all of the individual files on a single floppy disk, labels that disk “New York Times, October 31, 2000,” and sells copies of the disk to users as the electronic version of that day’s New York Times. The disk reproduces the creative, editorial selection of that edition of the New York Times. The reader, after all, has at his finger tips substantially all of the relevant content of the October 31 edition of the collective work. Moreover, each individual article makes explicit reference to that selection by including tags that remind the reader that it is a part of the New York Times for October 31, 2000. Such a disk might well constitute “that particular collective work”; it would surely qualify as a “revision” of the original collective work. Yet all the features identified as essential by the majority and by the respondents would still be lacking. An individual looking at one of the articles contained on the disk would still see none of the original formatting context and would still be unable to flip the page.

[54] Once one accepts the premise that a disk containing all the files from the October 31, 2000, New York Times can constitute a “revision,” there is no reason to treat any differently the same set of files, stored in a folder on the hard disk of a computer at the New York Times. Thus, at least before it is republished by the Electronic Databases, the collection of files that the New York Times transmits to them constitutes a revision, in electronic form, of a particular edition of the New York Times.

III

[55] The next question, then, is whether anything that the Electronic Databases do to the transmitted “revision” strips it of that status. The heart of the Court’s reasoning in this respect, as I understand it, is that, once received and processed by Electronic Databases, the data transmitted by the New York Times cannot be viewed as “revisions” within the meaning of §201(c) because of the way that data is stored and made available to the public by those Databases. First, the Court points to the fact that “the three Databases present articles to users clear of the context provided either by the original periodical editions or by any revision of those editions.” Ante, at 14. I have already addressed these formatting concerns. Second, and not wholly unrelated to the first point, however, the Court appears to think that the commingling of my hypothetical collection of ASCII article-files from the October 31, 2000, New York Times with similar collections of files from other editions of the New York Times (or from other periodicals) within one database would deprive that collection of revision status. See ibid. Even if my imaginary floppy disk could, in isolation, be considered a revision, the majority might say, that status would be lost if the floppy disk were to contain, not only the files from the October 31, 2000, New York Times, but also from the New York Times for every other day in 2000 (and other years) and from hundreds of other periodicals. I disagree.

[56] If my hypothetical October 31, 2000, floppy disk can be a revision, I do not see why the inclusion of other editions and other periodicals is any more significant than the placement of a single edition of the New York Times in a large public library or in a book store. Each individual file still reminds the reader that he is viewing “part of” a particular collective work. And the entire editorial content of that work still exists at the reader’s fingertips. [n13]

[57] It is true that, once the revision of the October 31, 2000, New York Times is surrounded by the additional content, it can be conceptualized as existing as part of an even larger collective work (e.g., the entire NEXIS database). See ante, at 14-15. The question then becomes whether this ability to conceive of a revision of a collective work as existing within a larger “collective work” changes the status of the original revision. Section 201(c)’s requirement that the article be published only as “part of … any revision of that collective work” does not compel any particular answer to that question. A microfilm of the New York Times for October 31, 2000, does not cease to be a revision of that individual collective work simply because it is stored on the same roll of film as other editions of the Times or on a library shelf containing hundreds of other microfilm periodicals. Nor does §201(c) compel the counterintuitive conclusion that the microfilm version of the Times would cease to be a revision simply because its publishers might choose to sell it on rolls of film that contained a year’s editions of both the New York Times and the Herald-Tribune. Similarly, the placement of our hypothetical electronic revision of the October 31, 2000, New York Times within a larger electronic database does nothing to alter either the nature of our original electronic revision or the relationship between that revision and the individual articles that exist as “part of ” it.

[58] Finally, the mere fact that an individual user may either view or print copies of individual articles stored on the Electronic Databases does not change the nature of the revisions contained within those databases. The same media-specific necessities that allow the publishers to store and make available the original collective work as a collection of individual digital files make it reasonable for the Electronic Databases to enable the user to download or print only those files in which the user has a particular interest. But this is no different from microfilm. Just as nothing intrinsic in the nature of microfilm dictates to a user how much or how little of a microform edition of the New York Times she must copy, nothing intrinsic in the Electronic Databases dictates to a user how much (or how little) of a particular edition of the New York Times to view or print. It is up to the user in each instance to decide whether to employ the publisher’s product in a manner that infringes either the publisher’s or the author’s copyright. And to the extent that the user’s decision to make a copy of a particular article violates the author’s copyright in that article, such infringing third-party behavior should not be attributed to the database. [n14] See Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417, 434 (1984).

IV

[59] My reading of “revision,” as encompassing products like the Electronic Databases, is not the only possible answer to the complex questions presented by this case. It is, nevertheless, one that is consistent with the statutory text and entirely faithful to the statute’s purposes. Respect for the policies motivating its enactment, to which I now turn, makes it wrong for the Court to reject this reading of §201(c).

[60] It is likely that the Congress that enacted the 1976 revision of the law of copyright did not anticipate the developments that occurred in the 1980s which gave rise to the practices challenged in this litigation. See Miller, Copyright Protection for Computer Programs, Databases, and Computer Generated Works: Is Anything New Since CONTU?, 106 Harv. L. Rev. 977, 979 (1993) (in 1976, “Congress … decided to avoid grappling with technological issues that obviously required more study than the legislative process was then willing to give them”). [n15] Thus, in resolving ambiguities in the relevant text of the statute, we should be mindful of the policies underlying copyright law.

[61] Macaulay wrote that copyright is “a tax on readers for the purpose of giving a bounty to writers.” T. Macaulay, Speeches on Copyright 11 (A. Thorndike ed. 1915) That tax restricts the dissemination of writings, but only insofar as necessary to encourage their production, the bounty’s basic objective. See U.S. Const., Art. I, §8, cl. 8. In other words, “[t]he primary purpose of copyright is not to reward the author, but is rather to secure ‘the general benefits derived by the public from the labors of authors.’ ” 1 M. Nimmer & D. Nimmer, Nimmer on Copyright §1.03[A] (2001) (quoting Fox Film Corp. v. Doyal, 286 U.S. 123, 127 (1932)); see also Breyer, The Uneasy Case for Copyright: A Study of Copyright in Books, Photocopies, and Computer Programs, 84 Harv. L. Rev. 281, 282 (1970) (discussing the twin goals of copyright law–protecting the reader’s desire for access to ideas and providing incentives for authors to produce them). The majority’s decision today unnecessarily subverts this fundamental goal of copyright law in favor of a narrow focus on “authorial rights.” Ante, at 21. Although the desire to protect such rights is certainly a laudable sentiment, [n16] copyright law demands that “private motivation must ultimately serve the cause of promoting broad public availability of literature, music, and the other arts.” Twentieth Century Music Corp. v. Aiken, 422 U.S. 151, 156 (1975) (emphasis added).

[62] The majority discounts the effect its decision will have on the availability of comprehensive digital databases, ante, at 19-21, but I am not as confident. As petitioners’ amici have persuasively argued, the difficulties of locating individual freelance authors and the potential of exposure to statutory damages may well have the effect of forcing electronic archives to purge freelance pieces from their databases. [n17] “The omission of these materials from electronic collections, for any reason on a large scale or even an occasional basis, undermines the principal benefits that electronic archives offer historians–efficiency, accuracy and comprehensiveness.”[n18] Brief for Ken Burns et al. as Amici Curiae 13.

[63] Moreover, it is far from clear that my position even deprives authors of much of anything (with the exception of perhaps the retrospective statutory damages that may well result from their victory today). [n19] Imagine, for example, that one of the contributions at issue in this case were a copyrighted version of John Keats’ Ode on a Grecian Urn, published on page 29 of our hypothetical October 31, 2000, New York Times. Even under my reading of §201(c), Keats retains valuable copyright protection. No matter how well received his ode might be, it is unlikely–although admittedly possible–that it could be marketed as a stand-alone work of art. The ode, however, would be an obvious candidate for inclusion in an anthology of works by romantic poets, in a collection of poems by the same author, or even in “a 400-page novel quoting a [poem] in passing,” ante, at 15. The author’s copyright would protect his right to compensation for any such use. Cf. Stewart v. Abend, 495 U.S. 207, 228 (1990) (discussing the value to authors of derivative works). Moreover, the value of the ode surely would be enhanced, not decreased, by the accessibility and readership of the October 31, 2000, edition of the New York Times. The ready availability of that edition, both at the time of its first publication and subsequently in libraries and electronic databases, would be a benefit, not an injury, to most authors. Keats would benefit from the poem’s continued availability to database users, by his identification as the author of the piece, and by the database’s indication of the fact that the poem first appeared in a prestigious periodical on a certain date. He would not care one whit whether the database indicated the formatting context of the page on which the poem appeared. What is overwhelmingly clear is that maximizing the readership of the ode would enhance the value of his remaining copyright uses.

[64] Nor is it clear that Keats will gain any prospective benefits from a victory in this case. As counsel for petitioners represented at oral argument, since 1995, the New York Times has required freelance authors to grant the Times “electronic rights” to articles. Tr. of Oral Arg. 7. And the inclusion of such a term has had no effect on the compensation authors receive. See ibid. This is understandable because, even if one accepts the majority’s characterization of the Electronic Databases as collections of freestanding articles, demand for databases like NEXIS probably does not reflect a “demand for a freelance article standing alone,” ante, at 11, to which the publishers are greedily helping themselves. Cf. Ryan v. Carl Corp., 23 F. Supp. 2d 1146, 1150-1151 (ND Cal. 1998) (“[T]he value added by the publisher to a reproduced article is significant”).

[65]  Instead, it seems far more likely that demand for the Electronic Databases reflects demand for a product that will provide a user with the means to quickly search through scores of complete periodicals. The comments of historian Douglas Brinkley are instructive in this respect:

“ ‘As an historian, when I want to write a biography, if I’m going to write a biography of Bill Clinton, the first thing I would do would be to index The New York Times. I would work through [the] microfiche and get any time Bill Clinton’s name ever appeared in The New York Times. I’d get a copy of that. So you’d have boxes of files. So for each month, here’s Clinton this month. You then would fill that in with … other obvious books or articles from Foreign Affairs or Foreign Policy or The New Yorker, or the like and you’d start getting your first biography of Bill Clinton.’ ” Panel Discussion: The Observer’s View (D. Brinkley, M. Frankel, H. Sidey), White House Historical Association (Nov. 16, 2000) (C-SPAN Archives No. 160577) (quoted in Brief for Ken Burns et al. as Amici Curiae 17).

Users like Douglas Brinkley do not go to NEXIS because it contains a score of individual articles by Jonathan Tasini. [n20] Rather, they go to NEXIS because it contains a comprehensive and easily searchable collection of (intact) periodicals. See id., at 8 (“The efficiency, accuracy, reliability, comprehensiveness and immediacy of access offered by searchable full-text digital archives are but a few of the benefits historians and other researches have reaped from the advancement in the technology of information”).

[66] Because it is likely that Congress did not consider the question raised by this case when drafting §201(c), because I think the District Court’s reading of that provision is reasonable and consistent with the statute’s purposes, and because the principal goals of copyright policy are better served by that reading, I would reverse the judgment of the Court of Appeals. The majority is correct that we cannot know in advance the effects of today’s decision on the comprehensiveness of electronic databases. We can be fairly certain, however, that it will provide little, if any, benefit to either authors or readers.