It’s all the same to me!: Copyright, contracts, and publisher self-archiving policies

Nancy Sims

* Editors’ note: From October 2015 through June 2016, two insightful colleagues (Maria Bonn and Nancy Sims) will share their thoughts while they navigate the ever-changing landscape of scholarly communication. Each will contribute two columns on different topics. This month’s column is Sims’s first, which sheds light on the copyright issues about different versions of a published scholarly work.
* Contact series editors Adrian Ho, director of digital scholarship at the University of Kentucky Libraries, and Patricia Hswe, digital content strategist at Penn State University, at E-mail: with article ideas.

“Green” open access—sharing copies of published scholarship online via repositories, rather than in the place of original publication—can be an appealing option for scholarly authors. It’s largely within their own control, and also often the option with least personal financial cost. Many publishers have standing policies enabling green open access of some kind, but the specifics of these policies vary widely and can be quite confusing for authors and others trying to understand and comply.

One of the major points of variation is which version of a published piece can be shared—the formatted version distributed by the publisher, the final manuscript subsequent to editing and review, or the manuscript version submitted by the author prior to review and editing. Early manuscripts are mostly of interest during the development of the piece, and are often not particularly relevant to the interests of either readers or authors once a piece has been edited and “finished.” Many readers may be quite happy with an open copy of the final edited manuscript, since it contains the same content as the finished article. But for authors, the final formatted publication may be preferred for a number of reasons.

For one thing, manuscript copies can be hard to locate. For another, it may be harder for subsequent users to cite manuscript copies in ways in which the citations are easily tracked back to the author. Since citation counts can be relevant to future employer and funder evaluations, this can be a fairly important consideration. These “version” distinctions have real effects as far as making open access “work” for authors and their potential readers, so it’s important to understand their origin and impact. There seems to be widespread belief among authors, library folks, and even some publisher representatives that these variations have to do with different copyrights in the various versions. This is almost entirely untrue. The distinctions are largely arbitrary.

“The copyright” in a work developed over time is a complicated concept. Under U.S. law (and the law of the majority of other countries), a copyright arises at the moment of creation of a work of authorship—in the United States, “[a] work is ‘created’ when it is fixed in a copy or phonorecord for the first time.”1

Practically speaking, though, many works (likely most scholarly works) are created through iterative processes involving multiple drafts, formal and informal review processes, significant editing, and revision. This usually takes months; sometimes years. The law does anticipate this kind of development over time—U.S. law says that “. . . where a work is prepared over a period of time, the portion of it that has been fixed at any particular time constitutes the work as of that time, and where the work has been prepared in different versions, each version constitutes a separate work.”2 But what does all this legal stuff mean in practice?

One thing is clear, there is no difference in “the copyright” in an unformatted manuscript and “the copyright” in that same text formatted for publication. In order to see why this is so clear, it may help to have a more robust understanding of some copyright theory: What gives rise to the existence of a copyright? Under U.S. law, copyright arises from the original expression of a creator and protects only that creator’s expression, not any facts or ideas that underlie the expression.3

Just a few of the many legal opinions that have explored these concepts can illustrate the basics. The widely persuasive opinion in Bridgeman Art Library v. Corel addressed whether a reproduction of an already existing work can have its own copyright. That court held that despite the work involved in creating photographic reproductions of paintings, there was no new expression in new photographs that were accurate reproductions, and thus no new copyright. Since the paintings in question in that case were in the public domain, “slavish copy” photos were also in the public domain.4

But even after the Bridgeman decision, other courts have found that reproductions of existing works that include new creative expression (such as a photograph of a sculpture that reflects expressive choices in angle, lighting, focus, etc.) can have a new copyright. These decisions are applying a much older Supreme Court case, Burrow-Giles v. Sarony. That case arose in the very early days of photography, when people disagreed about whether photographs, as images produced by machines, could be considered works of authorship in the sense needed to give rise to a copyright. The court ruled that photographs could be works of authorship eligible for copyright, due to expressive choices the photographer made in composing the image; posing the subject; choosing clothing, props, set dressing, and lighting; and so on.5 That’s one basic: new expression, new copyright; no new expression, no new copyright.

In U.S. law, effort alone is not enough to give rise to a copyright. It’s possible to invest quite a bit of effort and still produce something uncopyrightable. The most important decision on this point is Feist Publications v. Rural Telephone, in which the Supreme Court considered whether the work of collecting and compiling factual information could give rise to a copyright. They determined both that arrangements of factual information exhibiting “at least some minimal degree of creativity” could give rise to a copyright, but that a standard telephone directory did not reflect that level of creativity, however much effort was required to produce it.6

Sometimes, a creator of a work may invest a lot of effort to make many choices, and still not produce anything copyrightable. If the choices are not creative, but rather dictated by functional considerations, there is still not enough original expression for a new copyright. Some of the most interesting test cases on these points are specifically about shifting text content from one format to another: West Publishing has gone to court a couple of times to prevent copying of the way it reports U.S. court opinions. In no case has anyone ruled that West does not have a copyright in the new analysis it adds to court opinions—this is clearly original expression.

But West has at various times claimed that the choices it makes to systematize things, like attorney and court names, to standardize docket numbers and citations and their resulting pagination, are expressive choices that give rise to a copyright. This argument has been solidly rejected.7 The Second Circuit ruled that the only choices that give rise to a copyright are “subjective judgments relating to taste and value that were not obvious and that were not dictated by industry convention.”8 Thus, competitors (and public information websites) can copy all of the uncopyrightable material in a West-published court opinion, but they cannot copy the copyrightable analysis and annotations that West adds.

With that background, it’s clear that “the copyright” in a journal article or other scholarly publication arises from the creative, original, expressive choices made by the author (or authors) in the course of writing. Editorial and reviewer’s input may influence author choices, but still, the resulting copyrightable work is the text of the article. Preparing an article for publication may involve choices such as font size, column formatting, pagination, and so on, but those choices will usually be dictated by the standards of the journal and industry conventions, none of which are original expression that gives rise to a new or separate copyright.

But even with that background, another question remains. Is it possible that sometimes there are earlier versions of an article that are not affected by a copyright transfer agreement?

The law I quoted earlier, about works developed over time, mentions that the “portion of it that has been fixed at any particular time constitutes the work as of that time, and where the work has been prepared in different versions, each version constitutes a separate work.”9 Unpacking that statutory language a bit, it definitely does suggest that different versions of a work may have different copyrights. But it also suggests that there may be a unitary work that exists at different times—“the work as of that time.” One work may, over time, have different drafts that are basically the same work, or different drafts that are different works.

Most of the cases where courts consider whether two works are “the same” or “different” are cases where they are trying to determine whether one work infringes on another. Courts usually use the “substantial similarity” test to determine whether a later work infringes.10 There can be significant technicalities to the determination of substantial similarity, depending on the nature of the work (computer programs have multiple levels of comparison), and specific facts like whether the alleged infringer can be proven to have read (or seen, or heard, or used) the previous work. But ultimately the idea boils down to this: Where works are too similar, the later work contains too much expression present in the earlier work, so it infringes. Works that are not substantially similar contain separate copyrightable expression.

Cases about ownership of serial fiction are one of the few recurrent kinds of cases that address the internal differences of a work (or related works) created over time. A recent case about ownership of the Sherlock Holmes character reflects many of the baseline issues. In this case, the Conan Doyle estate argued that, since quite a few Holmes stories were published after 1923 (an important cut-off year for the public domain status in the United States), the standing copyright in those later stories gave them ownership and control over all uses of the character and related settings.

Instead, the Seventh Circuit affirmed a lower court’s ruling that any distinguishable elements, including characters, settings, and character traits, established in stories that are in the public domain, are free for anyone to use. Only the elements of the character newly established in later stories are covered by the later copyrights.11

One might conclude from the Sherlock Holmes case that the transfer of “the copyright” in an article only affects the parts of the article established at the very end of the development process—perhaps just the changes following review feedback. But keep in mind that the Holmes case was determining which elements of the characters were in the public domain, and which were added subsequently. The copyright in a character, and the copyright in a written work, are quite different things. In fact, some characters are not original enough to even have a standalone copyright. A character can, so to speak, stand on his or her own.

In the case of a scholarly work, elements that are severable from the overall narrative may function similarly to elements of a character. If an element can be separated from the rest of the written work, and does not appear in versions in which the copyright is transferred, the transfer of copyright in “the work” may not affect that element. But it is likely that transferring “the copyright” in a final edited manuscript also transfers copyright in any previous version that is substantially similar to that manuscript. It’s not easy to determine where the line of “substantial similarity” lies. It’s possible that early drafts of some works are so dissimilar to the final version that they have separate copyrights. But if an author has transferred away the copyright for a work they want to make more available, they probably won’t find it very satisfying to hear that the only copyright they still control is to the elements of the work entirely unlike the final version.

Whoever owns “the copyright” in an article or other scholarly work owns “the copyright” in both the final manuscript and the versions formatted for print or online publication. Authors who completely transfer their copyrights may retain the ability to distribute some versions of their work, but primarily via publisher policies that are subject to change at the publisher’s discretion. Any particular right of use that is important to an author should be explicitly addressed in the publication contract. Contracts provide expansive opportunities to create new legal constructs that meet the specific needs and wants of individual authors and publishers, which is convenient since “the copyright” in a work is a legal construct complex enough to make anyone’s head spin.

1. Title 17 of the United States Code, section 101—Definitions (entry for “created”):
2. Ibid.
3. Title 17 of the United States Code, section 102—Subject matter of copyright: In general:
4. Bridgeman Art Library v. Corel Corp, 36 F. Supp. 2d 191 (S.D.N.Y. 1999 ):
5. Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53 ( 1884 ):
6. Feist Publications, Inc., v. Rural Telephone Service Co, 499 U.S. 340 ( 1991 ):
7. Matthew Bender & Co. v. West Pub. Co, 158 F.3d 674 (2d Cir. 1998), cert denied, 526 U.S. 1154 ( 1999 ):
8. Ibid., 689.
9. Title 17 of the United States Code, section 101—Definitions (entry for “created”):
10. Arnstein v. Porter, 154 F.2d 464 (2d Cir. 1946):
11. Klinger v. Conan Doyle Estate, Ltd, No. 14-1128 (7th Cir. 2014):
Copyright © 2015 Nancy Sims

Article Views (Last 12 Months)

No data available

Contact ACRL for article usage statistics from 2010-April 2017.

Article Views (By Year/Month)

January: 7
February: 3
March: 2
April: 6
May: 2
June: 2
July: 3
August: 0
September: 1
October: 1
January: 2
February: 6
March: 0
April: 1
May: 3
June: 1
July: 3
August: 2
September: 2
October: 3
November: 1
December: 2
January: 3
February: 8
March: 0
April: 6
May: 0
June: 2
July: 3
August: 5
September: 1
October: 7
November: 0
December: 2
January: 10
February: 2
March: 4
April: 7
May: 3
June: 42
July: 8
August: 4
September: 7
October: 10
November: 2
December: 7
January: 13
February: 14
March: 6
April: 8
May: 8
June: 20
July: 19
August: 17
September: 4
October: 4
November: 17
December: 6
January: 7
February: 11
March: 46
April: 56
May: 81
June: 68
July: 11
August: 96
September: 13
October: 11
November: 5
December: 15
April: 2
May: 32
June: 4
July: 23
August: 88
September: 16
October: 6
November: 8
December: 10