discussed. Those issued have never before been mooted or determined by a court. In this case, an extensive trial was held before former Trial Judge James F. Davis who decided that the Government was liable for infringement. On review, helped by the briefs and agreements of the parties and the amici curiae we take the other position and hold the United States free of liability in the particular situation presented by this record.

1

The

Plaintiff, though a relatively small company, is a major publisher of medical journals and books. It publishes 37 journals, dealing with various medical specialties. The four journals in suit are Medicine, Journal of Immunology, Gastroenterology, and Pharmacological Reviews. Medicine is published by plaintiff for profit and for its own benefit. other three journals are published in conjunction with specialty medical societies which, by contract, share the journals' profits with plaintiff. The articles published in the journals stem from manuscripts submitted to plaintiff (or one of the medical societies) by physicians or other scientists engaged in medical research. The journals are widely disseminated throughout the United States (and the world) in libraries, schools, physiscians' offices, and the like. Annual subscription prices range from about $12 to $44; and, due to the esoteric nature of the journals' subject matter, the number of annual subscriptions is relatively small, ranging from about 3,100 (Pharmacological Reviews) to about 7,000 (Gastroenterology). Most of the revenue derived from the journals comes from subscription sales, though a small part comes from advertising. The journals are published with notice of copyright in plaintiff's name. The notice appears at the front of the journal and sometimes at the beginning of each article. After publication of each journal issue (usually monthly or bimonthly) and after compliance with the requisite statutory requirements, the Register of Copyrights issues to plaintiff certificates of copyright registration.

NIH, the Government's principal medical research organization, is a conglomerate of institutes located on a multi-acre campus at Bethesda, Maryland. Each institute is concerned with a particular medical specialty, and the institutes conduct their activities by way of both intramural research and grants-in-aid to private individuals and organizations. NIH employs over 12,000 persons--4,000 are science professionals and 2,000 have doctoral degrees. To assist its intramural programs, NIH maintains a technical library. The library houses about 150,000 volumes, of which about 30,000 are books and the balance scientific (principally medical) journals. The library is open to the public, but is used mostly by NIH in-house research personnel.

Demand by NIH research workers for access to plaintiff's journals (as well as other journals to which the library subscribes) is usually not met by in-house subscription copies. Consequently, as an integral part of its operation, the library runs a photocopy service for the benefit of its research staff. On request, a researcher can obtain a photocopy of an article from any of the journals in the library's collection. Usually, researchers request photocopies of articles to assist them in their on-going projects; sometimes photocopies are requested simply for background reading. The library does not monitor the reason for requests or the use to which the photocopies are put. The photocopies are not returned to the library; and the record shows that, in most instances, researchers keep them in their private files for future reference.

The library's policy is that, as a rule, only a single copy of a journal article will be made per request and each request is limited to about 40 to 50 pages, though exceptions may be, and have been, made in the case of long articles, upon approval of the Assistant Chief of the library branch. Also, as a general rule, requests for photocopying are limited to only a single article from a journal issue. Exceptions to this rule are routinely made, so long as substantially less than an entire journal is photocopied, i.e., less than about half of the journal. workers can, and frequently do, request single copies of the same article and such requests are honored.

Co

Four regularly assigned employees operate the NIH photocopy equipment. The equipment consists of microfilm cameras and Xerox copying machines. In 1970, the library photocopy budget was $86,000 and the library filled 85,744 requests for photocopies of journal articles (including plaintiff's journals), constituting about 930,000 pages. On the average, a journal article is 10 pages long, so that, in 1970, the library made about 93,000 photocopies of articles.

NLM, located on the Bethesda campus of NIH, was formerly the Armed Forces Medical Library. In 1956, Congress transferred the library from the Department of Defense to the Public Health Service (renaming it the National Library of Medicine), and declared its purpose to be "*** to aid the dissemination and exchange of scientific and other information important to the progress of medicine and to the public health ** *." 42 U. S. C. § 275 (1970). NLM is a repository of much of the world's medical literature, in essence a "librarians' library." As part of its operation, NLM cooperates with other libraries and like researchand-education-oriented institutions (both public and private) in a so-called "interlibrary loan" program. Upon request, NLM will loan to such institutions, for a limited time, books and other materials in its collection. In the case of journals, the "loans" usually take the form of photocopies of journal articles which

are supplied by NLM free of charge and on a no-return basis. NLM's "loan" policies are fashioned after the General Interlibrary Loan Code, which is a statement of self-imposed regulations to be followed by all libraries which cooperate in interlibrary loaning. The Code provides that each library upon request for a loan of materials, shall decide whether to loan the original or provide a photoduplicate. The Code notes that photoduplication of copyrighted materials may raise copyright infringement problems, particularly with regard to "photographing whole issues of periodicals or books with current copyrights, or in making multiple copies of a publication." [Emphasis in original text.] NLM, therefore, will provide only one photocopy of a particular article, per request, and will not photocopy on any given request an entire journal issue. Each photocopy reproduced by NLM contains a statement in the margin, "This is a single photostatic copy made by the National Library of Medicine for purposes of study or esearch in lieu of lending the original."

In recent years NLM's stated policy has been not to fill requests for copies of articles from any of 104 journals which are included in a so-called "widely-available list." Rather, the requester is furnished a copy of the "widely-available list" and the names of the regional medical libraries which are presumed to have the journals listed. Exceptions are sometimes made to the policy, particularly if the requester has been unsuccessful in obtaining the journal elsewhere. The four journals involved in this suit are listed on the "widely-available list." A rejection on the basis of the "widely-available list" is made only if the article requested was published during the preceding 5 years, but requests from Government libraries are not refused on the basis of the "widely-available list.'

Also, NLM's policy is not to honor an excessive number of requests from an individual or an institution. As a general rule, not more than 20 requests from an individual, or not more than 30 requests from an institution, within a month, will be honored. In 1968, NLM adopted the policy that no more than one article from a single journal issue, or three from a journal volume, would be copied. Prior to 1968, NLM had no express policy on copying limitations, but endeavored to prevent "excessive copying. Generally, requests for more than 50 pages of material will not be honored, though exceptions are sometimes made, particulary for Government institutions. Requests for more than one copy of a journal article are rejected, without exception. If NLM receives a request for more than one copy, a single copy will be furnished and the requester advised that it is NLM's policy to furnish only one copy.

In 1968.

a representative year, NLM received about 127,000 requests for interlibrary loans. Requests were received, for the most part, from other libraries or Government agencies. However, about 12 percent of the requests came from private or commercial organizations, particularly drug companies. Some requests were for books, in which event the book itself was loaned. Most requests were for journals or journal articles; and about 120,000 of the requests were filled by photocopying single articles from journals, including plaintiff's journals. Usually, the library seeking an interlibrary loan from NLM did so at the request of one of its patrons. If the "loan" was made by photocopy, the photocopy was given to the patron who was free to dispose of it as he wished. NLM made no effort to find out the ultimate use to which the photocopies were put; and there is no evidence that borrowing libraries kept the "loan" photocopies in their permanent collections for use by other patrons.

Defendant concedes that, within the pertinent accounting period, NLM and the NIH library made at least one photocopy of each of eight articles (designated by plaintiff as the Count I-to Count VIII articles) from one or more of the four journals in suit. These requests, as shown at the trial were made by NIH researchers and an Army medical officer (stationed in Japan) in connection with their professional work and were used solely for those purposes. In seven of the eight counts in the petition, the article requested was more than two years old; in the eighth instance it was 21 or 22 months old.

II

We assume, for the purposes of the case, but without deciding, that plaintiff is the proper copyright owner and entitled to sue here, and we agree with plaintiff that, on that assumption, it can sue for infringement of the eight separate articles." This faces us squarely with the issue of infringement.

Perhaps the main reason why determination of the question is so difficult is that the text of the Copyright Act of 1909, which governs the case, does not supply, by itself, a clear or satisfactory answer. Section 1 of the Act, 17 U.Ś.C. § 1, declares that the copyright owner "shall have the exclusive right: (a) To print, reprint, publish, copy, and vend the copyrighted work; ***." Read with blinders, this language might seem on its surface to be all-comprehensive-- especially the term "copy"--but we are convinced, for several reasons, that "copy" is not to be taken in its full literal sweep. In this intance, as in so many others in copyright, "[T]he statute is hardly unambiguous *** and presents problems of interpretation not solved by literal application of words as they are 'normally' used * * *." De Sylva v. Ballentine, 351 U.S. 570, 573 (1956). See, also, Fortnightly Corp. v. United Artists Television, Inc., 392 U.S. 390, 395-96 (1968).

***

The court-created doctrine of "fair use" (discussed in Part III, infra) is alone enough to demonstrate that Section 1 does not cover all copying (in the literal sense). Some forms of copying, at the very least of portions of a work, are universally deemed immune from liability, although the very words are reproduced in more than de minimis quantity. Furthermore, it is almost unanimously accepted that a scholar can make a handwritten copy of an entire copyrighted article for his own use, and in the era before photoduplication it was not uncommon (and not seriously questioned) that he could have his secretary make a typed copy for his personal use and files. These customary facts of copyright-life are among our givens. issue we now have is the complex one of whether photocopying, in the form done by NIH and NLM, should be accorded the same treatment-not the ministerial lexicographic task of deciding that photoduplication necessarily involves "copying" (as of course it does in dictionary terms).

The

One aspect of the history and structure of the 1909 Act offers another reason for refusing to give "copying" in Section 1, as applied to these articles, its simplest "ordinary" reach. It is pointed out to us, on the basis of analysis of the copyright laws from 1790 to 1909, that the early statutes distinguished "copying" from "printing," "reprinting," and "publishing," and provided that the copyright in books is infringed by "printing," "reprinting" and "publishing," while the copyright in other works (e.g., photographs, paintings, engraving, drawings, etc.) is infringed by "copying." Cf. Harper v. Shoppell, 26 F. 519, 520 (C.C.S.D.N.Y. 1886). The 1909 Act obliterated any such distinction in its text. It provides in §5 a list of all classes of copyrightable subject matter (including books and periodicals), and says in § 1 that the owner of copyright shall have the exclusive right "to print, reprint, publish, copy and vend the copyrighted work." Thus, the 1909 Act, unlike the earlier statutes, does not expressly say which of the prescribed acts of § 1 apply to which classes of copyrightable subject matter of § 5. Defendant and some of the amicisay that, to be consistent with the intent and purpose of earlier statutes, the "copying" proscription of 1 should not apply to books or periodicals; rather, only the proscribed acts of "printing," "reprinting" and "publishing" control books and periodicals. The proponents of this view stress that the legislative history of the 1909 legislation does not suggest any purpose to alter the previous coverage.

This is quite a serious argument. However, in view of Congress's general inclusion of the word "copy" in Section 1 and of the practice under the Act since 1909, we are not ready to accept fully this claim that infringement of periodical articles can come only through "printing," "reprinting" or "publishing.' But we do believe this point--that there is a solid doubt whether and how far "copy" applies to books and journals--must be taken into account in measuring the outlines of "copying" as it involves books and articles.