The opinion of the court was delivered by: SAND
Plaintiffs, affiliated commercial publishers of scientific journals, brought this action against defendants, not-for-profit physics societies. Plaintiffs seek to enjoin defendants from claiming in promotional materials that studies show defendants' physics publications, as measured by cost per printed character or cost per citation received within a particular timeframe, to be more "cost-effective" or "better bargains" than those published by plaintiffs. Plaintiffs claim primarily that the studies relied upon by defendants fail to measure cost-effectiveness, making the promotions literally false representations of fact, in violation of Section 43(a) of the Lanham Act, 15 U.S.C. § 1125(a). Some nine years after the original publication of the challenged material, and after parallel proceedings in three different foreign jurisdictions, two prior published opinions by this court, and a seven-day bench trial, the Court makes the following findings of of fact and reaches the following conclusions of law.
Plaintiffs OPA (Overseas Publishing Association) Amsterdam BV, Harwood Academic Publishers GMBH, and Gordon and Breach Science Publishers S.A. (collectively, "Gordon and Breach" or "G&B"), commonly known as the Gordon and Breach Publishing Group, are commercial publishers of over 300 journals in the fields of the physical and social sciences, arts and business. Among plaintiffs' publications are forty-six physics journals.
Defendant the American Institute of Physics ("AIP") is an umbrella organization of not-for-profit physics societies, including defendant the American Physical Society ("APS"). Both defendants publish physics journals; APS's 1996 revenues from publishing were $ 24 million, while AIP's 1995 publishing revenues were $ 38 million. Among defendants' publications are the Bulletin of the American Physical Society and AIP's Physics Today, a monthly magazine distributed to all members of AIP or its member societies.
A. The Barschall Articles
Beginning in the 1980's, academic journal subscription prices began to increase at a rapid pace. (Trial Tr. 751, 995; Pl. Ex. 77) As a result, libraries -- the primary subscribers to academic journals -- were forced to cut back on their subscriptions, with a twofold effect: competition within the scientific journal industry, in which the top 500 journals are a $ 2.3 billion-per-year business (Trial Tr. 756), became more aggressive, and a cycle of price increases ensued as publishers were compelled to recover publication costs from fewer subscribers, which in turn led to higher prices that caused more subscriptions to be dropped.
The conflict between the parties began in December 1986, when Henry H. Barschall, a physicist at the University of Wisconsin and a member of both an AIP publishing committee and the AIP Governing Board, published an article entitled "The Cost of Physics Journals" (the "1986 Article") in Physics Today. (Pl. Ex. 1) Beginning with the premise that "science libraries all over the United States face serious financial problems associated with the increased costs of journals," (id.), the 1986 Article reported a study in which Barschall used an averaging method to determine the number of characters per page of a journal, which was then multiplied by the number of pages published by the journal in that year. The resulting figure was divided into the journal's annual subscription price and multiplied by 1,000 to arrive at a cost-per-thousand-characters ("cost/kilocharacter") figure. Barschall then picked "at random one or two physics journals published by each of the major physics publishers," and presented them in tabular form sorted by cost/kilocharacter. (Id.)
Defendants' publications fared significantly better in the study than those of plaintiffs. The four journals published by defendants and listed in the table were the four highest-scoring journals, with costs ranging from 0.7 to 1.6 cents per thousand characters, while Particle Accelerators, the sole G&B physics journal listed in the table, was ranked last with a cost of 31 cents per thousand characters. While the article listed various factors which might account for the differences among some of the journals, it concluded by suggesting that libraries consider the low cost/kilocharacter of AIP and APS journals when making purchase decisions:
Following publication, AIP and APS distributed the study promotionally. Defendants mailed reprints of the article to librarians, timed to be received just before renewal bills for subscriptions to defendants' journals. APS's cover letter accompanying the reprint stated that, "using the proper quantitative measure -- the cost per published character (rather than the cost per annual subscription) -- Barschall demonstrates most convincingly that the journals of The American Physical Society (and the American Institute of Physics) are a great bargain compared to other physics journals." (Pl. Ex. 33) AIP's cover letter stated that Barschall's study was "based on an accurate quantitative measure: the cost per published character, rather than the cost per subscription. Barschall demonstrates that the journals of the American Physical Society and of the American Institute of Physics are a great bargain compared to other physics journals." (Pl. Ex. 34) Meanwhile, when it learned about the survey, Gordon and Breach wrote to Barschall and APS -- although Physics Today is an AIP publication -- complaining that there were various errors in the survey and requesting that G&B be eliminated from any future such surveys (Pl. Ex. 21). AIP disputes receiving the misaddressed letter (Havens Dep. 46-47), but the Court need not resolve the issue as nothing material turns upon it.
In 1988 Barschall undertook a more comprehensive survey (the "Barschall Study" or "Barschall's study") of over 200 physics journals, including 11 of the 24 physics journals then published by G&B. This survey measured journals not only by cost/kilocharacter, but also by "impact factor." The latter measure was taken from the 1986 Science Citation Index published by the Institute for Scientific Information ("ISI"), which defined a journal's impact as the average number of citations in 1986 to articles published in that journal in 1984 and 1985. For those journals for which an impact factor was available (ISI calculates an impact factor only for selected journals), Barschall also combined these two measures into a "cost/impact" ratio, which he characterized as "perhaps the most significant measure of the cost effectiveness of the journal." (Pl. Exs. 2 and 3). A full discussion of the study's methodology and results was published in the Bulletin of the American Physical Society, while an abbreviated version of the study, entitled "The Cost-Effectiveness of Physics Journals," appeared in Physics Today.
G&B's journals fared poorly in this study, as well. The Bulletin article listed all the journals included in the study in tabular form, sorted by cost/kilocharacter and cost/impact. AIP and APS journals scored well for the most part, with costs/kilocharacter ranging from 0.66 to 3.4 cents and costs/impact ranging from .063 to 2.8 cents. (Pl. Ex. 2) G&B's journals were clustered toward the bottom, with costs/kilocharacter between 16.0 and 31.0 cents and costs/impact ranging from 17.1 to 45.0 cents. (Id.) In terms of ranking, the 11 G&B journals were among the fourteen most expensive journals on a cost/kilocharacter basis, and the 4 G&B journals for which an impact factor was available were among the ten most expensive journals on a cost/impact basis. (Id.)
There is no simple solution [to the problem of increasing physics journal subscription prices], but authors can help physics libraries by publishing their papers in journals that have a low cost per character. In general, articles in such journals also have a greater 'impact,' so that authors too will benefit by publishing in them.
Defendants made a number of promotional uses of the new study. Before publication of the articles, AIP sent a letter to its members, stating that Barchall's just-completed study "shows that the APS publications are the greatest bargains around. Measured by the ratio of the number of citations to the cost of the journal, the APS publications score among the highest of all, with some of our competitors orders of magnitude lower." (Pl. Ex. 53) Defendants also distributed prepublication copies of the Bulletin article and Physics Today 's Table 2 to librarians. (Pl. Ex. 42) Following publication, defendants planned to distribute the Bulletin article to librarians, timed to coincide with the arrival of subscription renewal notices, with a cover letter which reproduced Physics Today 's table ranking publishers and stated that Barschall's study "places [AIP's] publications in the highest ranks of quality and value." (Pl. Exs. 65B, 71 at APS 003123) This mailing was cancelled after plaintiffs threatened to sue defendants. (Pl. Ex. 77 at APS 005165; Trial. Tr. 364-65) Defendants also made presentations to librarians, at which reproductions of Table 2 were shown. (Pl. Exs. 122, 130)
B. The Dispute Between the Parties
G&B wrote to AIP after publication of the 1988 articles, identifying various alleged errors in the Physics Today article and demanding a prompt retraction. (Pl. Ex. 61) AIP refused to issue a retraction, but published a statement in Physics Today reporting the dispute between the parties, informing readers that G&B had declined an offer to publish a letter in Physics Today stating its views, and inviting readers to contact G&B directly should they wish to learn more. (Pl. Ex. 74) In the same issue the magazine published a letter from Barschall which responded to some of G&B's complaints. (Id.)
Plaintiffs subsequently brought actions against defendants and Barschall in France, Germany, and Switzerland, and on September 23, 1993, filed the instant action. Plaintiffs alleged primarily that the Barschall studies constitute a literally false advertisement in violation of section 43(a) of the Lanham Act, 15 U.S.C. §§ 1125(a). Defendants disputed plaintiffs' claim and contended that, in any event, the doctrine of unclean hands prevented G&B from obtaining relief, because G&B had inequitably engaged in unjustified litigation and harassment as part of a program to suppress all criticism of its journals. In two prior published opinions which more fully explored the factual background of the case, this Court ruled, inter alia, that (i) claims regarding the 1986 Article were time-barred, Gordon and Breach Science Publishers S.A. v. American Inst. of Physics, 859 F. Supp. 1521 (S.D.N.Y. 1994) ("G&B I "); (ii) the original publication of the Barschall articles was protected speech under the First Amendment, but that their promotional distribution to the core consumers of physics journals fell within the scope of the Lanham Act, id. ; (iii) it would not give collateral estoppel effect to judgments favorable to the defendants rendered in the Swiss and German actions, Gordon and Breach Science Publishers S.A. v. American Inst. of Physics, 905 F. Supp. 169 (S.D.N.Y. 1995)
; and (iv) advertisements which touted defendants' journals with language such as "the most cost-effective prices," but which made no reference to the Barschall study, were mere puffery and thus fell outside the ambit of the Lanham Act. Id.
Plaintiffs' evidence at trial was directed toward demonstrating that librarians would be poorly-served if they relied exclusively on Barschall's study to determine journal cost-effectiveness. Plaintiffs relied primarily on two expert witnesses, Dr. Bruce Kingma, an economist at the State University of New York at Albany whose principal area of research is the economics of information and, in particular, library management (Trial Tr. 106-7), and Donald King, who has worked in information science for some 30 years. (Id. 416) While Kingma somewhat criticized the reliability of Barschall's methodology, the thrust of Kingma's testimony was that Barschall's study was an ineffective measure of cost-effectiveness, which embodies a claim both as to price paid and benefit received. Professor Kingma opined that Barschall's normalization of journals on a cost per character basis was inappropriate because a library's concern is not how much ink and paper it can get for its purchasing dollar, but how much use its patrons will make of the purchased material. (Id. 114). The impact factor does not compensate as a measure of use, stated Kingma, because different types of journals and different subdisciplines have different levels of citation, which he testified says nothing about their relative value or effectiveness. (Id. 117-24).
King concurred with Kingma that the most appropriate measure of cost-effectiveness was cost per use. (Id. 434) Like Kingma, he also felt that impact is a poor surrogate for use because it does not measure local use. (Id. 448) King also testified that cost per character is a poor basis for comparing journals because it fails to account for the justifications for such differentials (id. 433), primary among which was that niche journals, which are usually published by commercial publishers, serve a discipline having a limited audience so that the fixed costs of printing a journal must be borne by fewer subscribers resulting in higher costs per character. (Id. 437-41). As a result, he felt that a librarian relying only on a Barschall-type study would make "some horrendous mistakes." (Id. 440; 454, 491).
Defendants presented two experts to rebut plaintiffs' claims about the Barschall study's usefulness. Professor Norman Ramsey, professor emeritus of physics at Harvard University (id. 501), Nobel laureate in physics ( id. 504), and former chairman of the Governing Board of AIP and president of APS (id. 520), testified that libraries should not rely exclusively on the Barschall methodology in developing their physics journal collections, but that it was important for librarians to have the information contained in the Barschall study to take into account. (Id. 531-32). Michael A. Keller, University Librarian and Director of Academic Information Resources at Stanford University (id. 732), disputed Kingma's testimony that only local use and price should be considered in collection development; he testified that his library also considers the global practice of the discipline, other means for acquiring the information needed for a discipline, and future use of material. (Id. 736-37). He also testified that the identity of a journal's publisher may impact upon an acquisition decision (id. 740-41), and that normalization of journal prices on some quantitative basis is often performed. (Id. 747).
Another expert for defendants was Paul Ribbe, a mineralogist and professor emeritus at Virginia Tech University whose research focused on crystallography and the physics and chemistry of minerals. (Id. at 983-84). Among Ribbe's 100-plus published papers are four on bibliometrics (Def. Ex. U; Trial Tr. 989), and he has served as an academic journal editor and on his department's library committee. (Trial Tr. 985-86) Ribbe undertook a comprehensive quantitative analysis of physics journals, normalizing them on a number of unit cost and citation frequency bases. Ribbe's results, which revealed no marked differences among normalization bases, were proferred to demonstrate that Barschall's methodology yields reliable results.
In support of their unclean hands defense, defendants introduced evidence that G&B has engaged in an aggressive corporate practice of challenging any adverse commentary upon its journals, primarily through threatened (and actual) litigation. This evidence persuasively demonstrated that the present suit is but one battle in a "global campaign by G&B to suppress all adverse comment upon its journals." (Def. Post-Trial Mem. at 17-18) Although ultimately it is unnecessary to reach defendants' affirmative defense, the Court sets forth some of the facts on which the defense is based because they ...