I’m going to turn this over to Betsy Rosenblatt, current colleague/head of the legal committee of the Organization for Transformative Works and Assistant Professor of Law/Director, Center for Intellectual Property Law at Whittier Law. Professor Rosenblatt is also the author of an article about Holmes-related IP issues, so she’s the best possible commentator.
Arthur Conan Doyle’s Sherlock Holmes canon, comprising 56 short stories and 4 novels, are among the most read and most-adapted works in recent history. In most of the world, copyright in all of the stories and novels has expired. In the U.S., copyright protection has expired on some of the stories but, thanks to the Copyright Term Extension Act, copyright protection remains for ten stories, first published after 1923 in the U.S. under the collective title “The Case-Book of Sherlock Holmes.” These rights belong to a company called the Conan Doyle Estate (“CDE”), which has licensed them to a number of adapters, including Warner Brothers (for the Robert Downey Jr. Sherlock Holmes films), CBS-TV (for Elementary), and the BBC (for its distribution of its Sherlock series in the U.S.).
On December 23, 2013, Judge Ruben Castillo of the Northern District of Illinois ruled in the case of Klinger v. Conan Doyle Estate, Ltd. that U.S. copyright protection had expired on all elements of the Sherlock Holmes canon that were first introduced before 1923. That part of the case is straightforward, and no doubt correct.
But the case also has broader implications worth exploring. Most notably, while several courts have held that “increments of expression” original to derivative works may be protectable, they’ve provided little to no guidance about what constitutes a protectable “increment of expression” about a copyrightable character. This ruling sets the bar uncomfortably low—implying that any new fact or trait about an already distinctly delineated character may create a new, protectable version of that character, no matter how generic, predictable, or insignificant that trait.
1. Copyright in serialized characters
Klinger structured his complaint very narrowly to focus on copyright expiration. He could have asked whether his anthology or similar commercial derivative works constituted copyright fair use. He could have asked for a declaration concerning the CDE’s claim (made elsewhere) that it has trademark rights in the word mark “Sherlock Holmes” for a number of classes of goods, including detective fiction. As it was, Klinger asked only for a declaration concerning the copyright status of the pre-1923 story elements and post-1923 stories.
This narrow framework allowed the court to rule on summary judgment, reaching the generally self-evident conclusion that story elements first published prior to 1923 are no longer protected by copyright, and those protectable elements introduced later are still protected. This is a straightforward and wholly reasonable result for the characters, plots, and story elements that appeared exclusively in pre-1923 works. But what of characters that appeared in both pre-1923 and post-1923 works?
The CDE offered what the court described as a “novel legal argument” that the characters of Sherlock Holmes and Dr. Watson continued to be developed throughout the post-1923 stories, and that therefore, the characters themselves would not enter the public domain until their last story did. The court rejected this argument, explaining—in line with Silverman v. CBS and a number of other cases that have held similarly—that “where an author has used the same character in a series of works, some of which are in the public domain, the public is free to copy [character and] story elements from the public domain works,” but not from still-protected works. The same rule would apply broadly to not only Holmes and Watson, but also other characters created early in the 20th century and serialized over time, like G.K. Chesterton's Father Brown, Edgar Rice Burroughs' Tarzan, and Agatha Christie's Hercule Poirot.
Although the court explicitly avoided deciding “the copyright status of the Sherlock Holmes character” as a whole (fn 8), as a practical matter, the result of this opinion is that the characters of Holmes and Watson are in the public domain. To some extent, this result is an artifact of the manner in which Klinger sought relief: in an exhibit to his complaint, he listed all of the story elements that he argued were on the public domain, including a number of complete characters (such as Inspector Lestrade, Mrs. Hudson, and Irene Adler), and “as to Holmes himself” and “as to Dr. Watson,” a number of character traits. The court therefore was never asked to, and did not, perform an explicit analysis of whether Holmes, Watson, or other characters reached the threshold of “distinctly delineated” characters worthy of copyright protection—but the clear implication of this ruling is (a) that Holmes and Watson, as characters, were sufficiently distinctly delineated in the pre-1923 stories to be protected by copyright, and (b) that that copyright has expired. To the extent that the characters were modified in the ten post-1923 stories, those modified versions of the characters remain protected. The CDE protested that this holding would “dismantle” the characters of Holmes and Watson into a public domain version and a copyrighted version. The court was unconcerned.
Personally, I’m less worried by the idea of “dismantled” characters than by the particular character elements that the court found protectable. The court identified three character traits of Holmes and Watson as still protected: (1) Dr. Watson’s second marriage; (2) Dr. Watson’s background as an athlete; and (3) Sherlock Holmes’s retirement from his detective agency. As a factual matter, this result is also an artifact of the case’s procedural history—the court only considered the parties’ factual submissions and didn’t review the text of Arthur Conan Doyle’s actual canonical writings. If it had, it would have observed, for example, that Holmes’s retirement was actually first described in detail in a public domain story, “His Last Bow.” Likewise, Watson’s second wife is factually far from the “character” the court may have assumed she was—she is mentioned in passing and never even given a name, much less identified by “distinctly delineated” character traits.
Setting these factual matters aside, however, the ruling raises a deeper question: what does it mean for a public domain character to have a protected character trait? Certainly, the court would not protect the generic ideas of “Doctor with a background as an athlete” or “Doctor with a second wife.” And the court has made clear that Watson himself has fallen into the public domain—but his history as an athlete and the existence of his unnamed second wife have not. The court reasoned that the later-published stories constituted derivative works from the public-domain stories, and provided “enough expressive variation from public-domain or other existing works to enable the new work”—here, the historically athletic, second-married version of Watson—“from its predecessors.”
On one hand, this conclusion likely differs from the way most people experience serialized characters. We see them as unitary entities, about whom we simply learn more over time as more facts are revealed to us. We likely imagine that those facts have always been there, even if we did not know them. This is particularly true of characters like Holmes and Watson, since their tales were not published in chronological order. We therefore experience their lives out of sequence, and assign the sequence as we read. But dividing characters into multiple versions—Watson 1.0 was distinctly delineated before 1923, and Watson 2.0 came later—reflects the reality of serialized production and, more importantly, permits characters to enter the public domain in an orderly manner. The alternative would be far worse: copyright holders could, in effect, create eternal copyright in characters simply by continuing to write new stories about them.
But while the idea that new versions of characters may retain protection after old versions fall into the public domain may be a fine one, the ruling highlights a line-drawing problem: how new does a “new character trait” have to be to create a new version of an old character? Drawing on case law about the level of originality required for copyrightability in general, this court set the threshold as low as possible: all that is required is an “original expression” sufficient to “enable the new work to be readily distinguished from its predecessors.” Although this is consistent with the law about protecting original “increments of expression” in derivative works, it’s an odd result as applied to characters, since courts have generally required a much higher level of originality—“distinct delineation”—for copyright to attach to a particular character. The court notes that a “character, character trait, and storyline” may each be copyrightable increments of expression in a derivative work. But why not require the same degree of distinctness from prior works to protect Watson 2.0 as to protect Watson 1.0 in the first instance? The court does not explain.
Under this court’s reasoning, once a creator has gone to the trouble of distinctly delineating a character, a relatively small increment of new information about that character would, in essence, always create a “new” version of the character, protectable until expiration of the new copyright. This isn’t a significant problem for adapters of Conan Doyle’s Holmes canon, since nearly every meaningful trait about the characters and their settings was established in the pre-1923 stories—in essence, therefore, the characters are entirely in the public domain, but for three (indeed, arguably two) character traits. It’s also not a problem for new works of individual authorship, since copyright protection for such works is tied to the date of the author’s death, rather than the date of publication. But the reasoning of this ruling could create a way for owners of work-for-hire copyrighted characters to, effectively, extend protection indefinitely by making incremental changes to their characters over time. That’s worrisome.
2. Declaratory judgment jurisdiction
The opinion also addressed declaratory judgment jurisdiction in a post-DMCA world. The CDE argued that the court should decline jurisdiction because there was no “actual controversy” between the parties—Klinger had not yet produced his book, and had no reasonable apprehension of litigation because the CDE hadn’t actually threatened to sue. Its only explicit threat had been to use the DMCA and keep Klinger’s book out of major retailers such as Amazon and Barnes & Noble. The court rejected the CDE’s argument, ruling that, per Medimmune, the actual controversy requirement may be satisfied without a threat of litigation, and that the CDE’s DMCA threat established the existence of clear, adverse legal interests. The CDE also argued that the court should decline to exercise jurisdiction for prudential reasons, as the court had no access to Klinger’s proposed book, but the court reasoned that it could resolve the entire controversy by resolving the copyright status of the elements Klinger proposed to include in the book.
This result seems right, for a number of reasons. First, it recognizes the business realities of publishing: threatening to keep a book out of major retailers (and thus out of consumers’ hands) is, no doubt, tantamount to threatening litigation. It also helps reinforce the important idea that cease & desist recipients have alternatives other than backing down. More importantly, it can help alleviate the power imbalance set up by the DMCA, whereby rights holders can have works taken down with relative ease, but having them put back up is much more difficult. Of course, declaratory jurisdiction is far from a panacea for that power imbalance—bringing a lawsuit is even more daunting than sending a DMCA counter-notification!—but in situations like this, where the DMCA claimant is relying on dubious rights, it provides an opportunity for something broader than case-by-case counter-notification.
3. Impact on future adapters
This case draws a relatively clean line around the CDE’s copyrights, such as they are, and therefore will likely provide significant freedom to adapt the original Conan Doyle Holmes canon. This is because the preclusive impact of this case is uneven, in adapters’ favor: as a matter of civil procedure (specifically, the doctrine of non-mutual defensive issue preclusion), the present ruling will likely preventing the CDE from prevailing in pre-1923-based copyright claims not only against Klinger, but also against other Holmes adapters—but it will probably not (according to the doctrine of non-mutual offensive issue preclusion) bind future courts to this ruling’s three examples of post-1923 protected material. This will undoubtedly make it more difficult for the CDE to assert claims against adaptations not specifically based on post-1923 works. And of course even those adaptations may constitute fair uses—in fact, it’s entirely likely that Klinger’s proposed anthology would have been a fair use even if the characters of Holmes and Watson had not been in the copyright public domain—but fair use questions will have to wait for future cases.
It also won’t prevent Holmes adapters from suing each other. In fact, most cases about intellectual property in Sherlock Holmes have been among adapters: for example, Granada v.Lorindy involved claims among two producers of Sherlock Holmes television series, and Musto v.Meyer involved a claim by an author of a medical journal article that the book and film “The Seven Per Cent Solution” infringed copyright in his article. This case may be informative in such cases, but will not bind the courts.
Trademark questions must also wait for another day. The CDE’s attorney told the New York Times the case has no impact on its “existing trademark claims.” This refers to the fact that the CDE has actively pursued trademark protection for a number of words and images relating to Sherlock Holmes, including the CDE’s logo and the words “Sherlock Holmes.” This case has no direct impact at all on any of the CDE’s trademark assertions, although it does highlight the fact that that many different entities have used, and will continue to use, the names and characters of Holmes and Watson without permission from the CDE—a fact that significantly undermines any claim by the CDE that the characters or their names have “secondary meaning” as source identifiers for the CDE.
RT’s comments: this case is a no-brainer, and in fact is evidence of just how crazily expansive copyright is: I agree that the loose language in the opinion suggesting that the incremental expression in “Holmes retired” or “Watson remarried” is protectable is quite disturbing. At the very least, the doctrine of merger should prevent any control over the basic postulate that Holmes retired, even if the details in Doyle’s in-copyright story must be avoided in the absence of fair use or some other exception. As the court itself notes, the increment that has to be added for copyrightability is an increment of expression, not an increment of just anything. It doesn’t make sense to say, as the court does, that “characters, character traits and storylines” are inherently increments of expression as opposed to “ideas, plots, dramatic situations, and events,” which are not. (What’s the difference between a storyline and a dramatic situation or a plot? What about some dramatists’ contention that all plot derives from character?) I’m heartened, however, by the court’s caution that neither party submitted any portion of the canon for review. This is a ruling on a declaratory judgment plaintiff’s motion for summary judgment in the absence of evidence.
As for the trademark issues Professor Rosenblatt highlights, they too should be no-brainers under Dastar, and I hope and trust that courts will see this. In fact, I suspect the Holmes ouvre might be a particularly good test case for preserving the freedom to copy, and to tell people what you’ve copied.