In 1995, I wrote an article for the Chronicle of Higher Education outlining the problems of publishing scholarly books in literary criticism and explaining why the Penn State University Press could no longer afford to remain active in this field. Of the 150 books about literature the Press had put out in the previous decade, 65% had sold fewer than 500 copies, 91% fewer than 800 copies, and only 3% more than 1,000. The pattern of sales in this discipline had eroded to the point where a press without much of a subsidy from its parent university could not sustain a publishing program in it anymore. It seemed clear even then that what we scholarly publishers have come to call the problem of “endangered species” would be spreading to other disciplines over time. Five years later, in an article I wrote for the newsletter of APSA's Organized Section on Comparative Politics (2000), I analyzed data that seemed to show that field to be heading in the same direction as literary studies, and I concluded with not a great deal of hope for the future. Recently, at the invitation of the Association for Political Theory, I turned my attention to the subfield of political theory and offered this paper as background for the session on book publishing at the conference in November 2006. While many of the same pressures remain in place to bedevil university presses, and it would be premature surely to claim that we are out of the woods yet, there have been some significant changes that give reason to think the future may not be quite as gloomy as it appeared back at the turn of the millennium.

First, before talking about the recent changes, let's look at some numbers that illustrate how the market has eroded over the past few decades. My previous employer, Princeton University Press, did detailed studies of sales patterns in different disciplines. For political science as a whole, the average five-year total for books published only in hardback in the period 1960–1967 was 3,387. That average had already dropped to 1,768 for books published between 1971 and 1973 and by 1979–1981 was down to 1,274. Paperbacks began emerging in the late 1960s, but initially were typically released after the hardback had already been out for at least a couple of years. In political science, the average five-year sale for later paperbacks at Princeton was 2,623 for paperbacks published in 1973–1977. Helped by an NSF-sponsored study (Fry and White 1975) of changing library budgets (documenting a trend of more acquisitions funds going to journals and less to monographs from 1969 to 1973), university presses began recognizing the seriousness of the erosion in library sales as early as the first half of the 1970s and adopted a new strategy of trying to recoup some of the lost hardback sales by issuing more titles simultaneously in hardback and paperback. By 1980, every book in political science at Princeton was being issued simultaneously in hardback and paperback. The average five-year sale for hardbacks in dual editions was 1,206 in 1977–1979 but only 996 by 1981–1983, while the paperback averages for those periods were 3,754 and 5,481, respectively. On my recommendation, Princeton began tracking sales of political philosophy books separately in the later 1970s. For books published in hardback only during these two periods, the average five-year sale actually increased from 1,292 to 1,440 (but the number of titles included were very small, fewer than five). For books issued simultaneously in hardback and paperback, the averages were 1,120 down to 975 for hardbacks and 4,393 up to 5,267 for paperbacks.

With this relatively encouraging experience as background, I carried over the same strategy to Penn State when I came here in 1989. Many other presses by that time had long since jumped on the bandwagon of dual editions that Princeton had pioneered in the early 1970s with its simultaneous Limited Paperback Editions (LPEs), and competition with other presses, not to mention expectations from authors, made this the dominant approach in most fields of scholarly publishing, especially in the social sciences but also in many fields of the humanities, too. But the numbers were already beginning to suggest that this strategy had its limits. Let me illustrate by using data from sales of the 73 books in political theory that Penn State published over the 15-year period from 1991 to 2006. For convenience I'll group them into clusters, dividing them into three five-year periods during which the Press published 29, 24, and 20 titles, respectively (reflecting the overall pattern of initial growth of the Press's annual output to a maximum of 80 titles in the mid-1990s and then a gradual retrenchment to about 50 titles a year currently).

In the first two periods, 1991–1996 and 1996–2000, the Press published only nine titles that were not dual editions and 44 that were, and of the nine, two were paperback reprints of books by Stephen Bronner and Jean Bethke Elshtain that commercial publishers had allowed to go out of print, leaving just seven as books issued only in hardback. Excluding Chris Sciabarra's atypical Ayn Rand: The Russian Radical (1995), which enjoyed significant book club sales and total sales of 2,530 hardbacks and 7,385 paperbacks, the average total sales through June 2006 for the 43 remaining titles were 466 hardbacks and 1,366 paperbacks for the 23 books published in 1991–1995 and 243 hardbacks and 931 paperbacks for the 20 books published in 1996–2000. These represent declines of 48% and 32%, respectively—sobering numbers for any publisher.

We knew already from statistics issued by the Association of Research Libraries that since the mid-1980s academic library purchases of monographs had dropped nearly 25% as an ever-greater share of their funds had gone toward sustaining journal subscriptions (even after libraries began cancelling subscriptions in the early 1990s). Anecdotal evidence, as well as information gleaned from Yankee Book Peddler (the largest wholesale supplier of academic books to libraries), suggested that more libraries than ever were opting to buy paperback editions instead of hardbacks when they were issued at the same time, thus contributing further to the drop in cloth sales that are so vital to the economics of scholarly publishing. Meanwhile, the proliferation of coursepacks in the early 1990s and the increasing resort to e-reserve and course-management systems (like Blackboard and WebCT) as they came to be offered later in the decade, tempting more teachers to cannibalize books for excerpts rather than assign whole paperbacks to their students (and often not even paying permission fees for such reproduction), began cutting into the income presses had been deriving from that side of the market. As the new century dawned, some presses, including Penn State, responded to these new challenges by retreating to the earlier paradigm of initial hardback publication followed by a paperback edition a year or two later. While the sales of the paperback overall may not reach the same level (unless a book happens to become a staple of adoption for courses), the chance to sell 350–500 copies of a hardback in the first 12–24 months instead of 200 or fewer is persuading more presses that this is the only viable option left, short of abandoning publication in the field altogether. Our experience at Penn State so far has borne this out. Reversing the earlier trend, in 2001–2006 we published 20 books in political theory, only seven of which were done as dual editions (two were books in Nancy Tuana's “Re-Reading the Canon” series, where we had established this pattern for the series long ago, and another one was a translation of a book by German theorist Otfried Höffe). Of the 13 titles published initially in hardback only, the average sale to date has been 436 hardbacks, with the average sale of the 10 later paperbacks that have been out for at least a year being 178 copies. This compares with the average sale for the seven dual-edition titles of 181 hardbacks and 431 paperbacks. Interestingly, the difference between the hardback and paperback sales is almost exactly the same, but the extra 255 hardback copies sold, at prices usually two to three times higher than the paperback prices, make a substantial difference in the bottom line. For the time being, then, this reversion to the older approach seems to be working, and authors generally do not mind the delay in the release of a paperback, which (excluding one title where the paperback followed the hardback over three years later and another where the paperback has not been released yet for a title just published in February) has averaged 16.9 months for the 11 books issued in hardback only during the 2001–2006 period. Advantages of publishing the paperback later include, for the author, a jacketed hardback edition and an additional round of advertising upon publication of the paperback, which might also be able to include quotes from reviews on the back cover by that time.

Other options exist, of course. One is to adopt the business model long followed by European publishers: print only a few hundred copies of a new book in hardback, price it high enough to recover all costs from sales to libraries (say, $125 for a 200-page monograph), and then any sales to individuals become icing on the cake. U.S. presses have shied away from this approach, probably because we feel it is part of our basic mission as scholarly publishers to distribute as many copies as we can at as low a price as possible while still remaining solvent. Another option is to seek more subsidies from the universities whose faculty become our authors. Some universities are becoming more aware of the need to provide this kind of support, especially for junior faculty, as pressures for publishing monographs continue while outlets for them dry up. Stanford, for example, announced a couple of years ago that it would grant $5,000 each to all junior faculty members for use in whatever way they best deemed to advance their careers, including as title subsidies to publishers. But why, then, shouldn't universities just subsidize their own presses more? We press directors would welcome that option, but it isn't being offered by many universities. A recent report (2006) from the Association of American University Presses (AAUP) shows that of the 63 presses participating in its survey for 2002–2005, only 49 received any operating subsidies, and the increase in parent-institution support was 8% over this period, just barely keeping up with the period's 7% rate of inflation. At any rate, it seems unfair that the universities that do have presses should bear the full burden of sustaining the system of scholarly communication on their own when it benefits scholars from all institutions; in this light, clearly title subsidies are a more equitable solution, as long ago recommended by the American Council of Learned Societies in their National Enquiry into Scholarly Communication (1979).

But what about saving costs by making greater use of technology? In fact, presses have been taking advantage of advances in technology for at least three decades, and the savings—in copyediting, design, marketing, order fulfillment, etc.—have been sufficient to keep the rate of increase in prices far below what it would otherwise have been in the face of steadily eroding sales. More recently, there has been much talk not just about deploying technology to enhance efficiencies in producing books in the traditional manner but also about doing actual electronic publishing. Are e-books the wave of the future? An initial wave of enthusiasm for e-books lured a number of large commercial publishers into setting up electronic publishing divisions in the heady times of the dot.com era of the late 1990s and early 2000s. But the hopes for an early booming market never materialized, in part because of the competition among a variety of proprietary e-book reader vendors whose systems were all incompatible with one another, but also in part because the initial focus was on trade books, which proved to be the least amenable to e-book innovation. Most of these electronic publishing divisions were closed down, having burned through lots of money with few results to show for the effort. More recently, signs of a more robust if not booming market for e-books have emerged, and with the advent of Sony's new e-book reader and of other sophisticated technologies deploying electronic ink, hopes are again high that e-books will eventually grow into a major piece of the market. Meanwhile, companies like eBrary, netLibrary, and Questia, which managed to survive the dot.com massacre (some just barely), have provided opportunities for publishers to have their backlists of pre-2000 titles digitized and sold in electronic form. And, beginning in 1999, Lightning Source, a subsidiary of the major book wholesaler Ingram, approached publishers to store their titles in electronic form so that print copies can be produced “on demand” (even one at a time), thus eliminating the need for publishers to keep inventory of slow-selling books.

The emergence of online retailers like Amazon.com and Barnesandnoble.com also proved initially very helpful to publishers, as they enabled much wider access for readers to backlist titles and, with Amazon's “Search Inside the Book” feature, the ability to browse before buying. (A more recent initiative of online retailers, to sell used books, has proved detrimental to publishers, however.) With these innovations already giving new life to what has come to be known as the “long tail” of the market, the entry of Google with its Book Search and Google Scholar features, as well as its AdWords program, has made this an even more robust market. It remains to be seen how significant financially these technological developments will be in providing new streams of income to offset the declining stream from the traditional marketing of books through direct-mail catalogues, space advertising, and “bricks and mortar” stores. What can be said with some confidence now, after years of trial and error, is that those publishers who have experimented with doing full electronic publishing have quickly discovered that the added hardware, software, and staff costs of making the technology work equal, if not exceed, the traditional costs of typesetting, printing, binding, and warehousing—which themselves only amount to about 40% of the overall expense of publishing a book. Thus, e-publishing offers no panacea for the dilemmas faced by scholarly book publishers today (though it has already proven to have much greater benefits for journal publishers).

Where technology has had undoubtedly the most direct and immediate impact on scholarly publishers' finances is in the advent of digital printing as an alternative to traditional offset printing. The constraints of the latter technology led publishers constantly to overprint, in the pursuit of lower unit costs, and resulted in a huge amount of waste as well as unnecessary capital investment in inventory, much of which later had to be destroyed. With digital printing, supply can be more finely tuned to actual demand with low initial print runs and frequent reprintings in small quantities, which eliminate the need to keep substantial inventory and thus significantly increases cash flow. The effect on scholarly publishers' bottom lines has been nothing short of liberating, and together with the pure print on demand (POD) offered by Lightning Source for lower-selling titles, it has contributed more to the financial health of the industry than any other innovation in recent decades.

This kind of technological revolution will surely help us keep our heads above water for a while, and to this extent it may stave off thrusting political theory into the category of “endangered species,” at least for the near future. Presses are still coping with numerous pressures, however, and these lead generally in the direction of publishing fewer scholarly monographs. Another statistic from the AAUP report begins to loom large here: collective title output from 63 university presses was almost exactly the same in 2005 as in 2002 (and actually lower in 2003 and 2004), and had been flat in the period 1998–2001 as well, whereas in the 1980s output was still increasing at double-digit rates and even as late as 1995 was up 10% over the previous year. With increasing pressures on presses to pay their own way, many have resorted to changing the “mix” of what they publish, substituting more saleable titles (like regional books, reference works, “trade” titles, even textbooks) for less saleable monographs. At Penn State we have cut our annual output by about 35% in the past decade, and those cuts have mostly come in traditional monographic studies, many of which haven't even sold well enough to recoup their manufacturing costs, let alone contribute anything to overhead; the time and money we have saved have been devoted to getting “more bang for the buck” out of titles with greater sales potential, especially regional books. Political theory, because of its sales profile, has been one of the areas in which we have reduced our output, from an average of 5.8 titles annually in 1991–1995, to 4.0 titles in 2001–2006. I suspect a number of other presses have done the same.

As if this were not bad news enough for all scholars in the field, there is even more bad news for junior faculty. More universities (including Penn State) in the past few years have joined the movement associated with the Networked Digital Library of Theses and Dissertations (NDLTD), founded at Virginia Tech in the mid-1990s, which has the noble goal of making dissertations in the future more readily accessible worldwide for scholarly use through the Internet. Meanwhile, ProQuest (formerly UMI) has encouraged submission of dissertations in electronic form, and a number of universities (like Penn State) have mandated that all dissertations be submitted as eTDs in the future. With graduate students having few options to opt out of this kind of system, they may find that as junior faculty they have limited their opportunities for publishing revised versions of their dissertations as their first books. Libraries, which can readily tap into the NDLTD network and can subscribe to the ProQuest database, may think twice about spending scarce resources on books when they know that they can access the original dissertations either through the NDLTD or ProQuest. In fact, as reported by Yankee Book Peddler, there are a substantial number of libraries that instruct the vendors who handle their approval plans to inspect newly published books to find out if they started as dissertations and, if they did, to exclude them from their orders. At the very least, the desirability of substantially revising a dissertation to turn it into a book has now become a necessity, and junior faculty should not be surprised at being queried on the nature and extent of revision before an editor will invite submission of their manuscripts. I issued a warning to this effect in a letter-to-the-editor of PS: Political Science & Politics (2006), the APSA's main medium for communicating with its membership.

On the brighter side is experimentation with new kinds of “enhanced” books, which takes maximum advantage of new technology. This is already being pursued by the American Historical Association (AHA), with support from the Mellon Foundation, in the Gutenberg-e Project. Robert Darnton, when AHA president, elaborated this vision in his article (1999) on “The New Age of the Book.” It is also being carried out in the ACLS History E-Book Project, likewise initially supported by the Mellon Foundation. With no great constraints on space in the electronic world, e-books in history, humanities, and the social sciences can incorporate data sets, interview materials, documentary appendices, hyperlinks to other works cited and to digital archives, digitized maps, color illustrations, even audio files, to create much more comprehensive, multimedia publications. They can be constructed in “layers” aimed at different audiences. And there can be opportunities for readers to engage in constructive feedback online, making their responses part of a work growing in complexity over time as a larger collaborative enterprise, much in the way that the Wikipedia works today. The mind boggles at the possibilities! But none of this is cheap—and it certainly won't solve the immediate problem of the “endangered” traditional monograph anytime soon. Nor will the recent calls for making everything “open access,” which we hear from the provosts supporting the Federal Research Public Access Act and from the ACLS Commission on Cyberinfrastructure for the Humanities and Social Sciences. Theirs is a noble vision, which we presses share as an ultimate goal, but they unfortunately do not offer any specifics on how to construct the new business models that their vision depends upon to be realized. Wishing will not make it happen, alas.

For the immediate future an experiment we are undertaking at Penn State may be more relevant, and it does involve a form of open access. Under the auspices of our Office of Digital Scholarly Publishing, which the Press launched jointly with the University's libraries in the spring of 2005 before becoming administratively a part of the libraries in December, we are coming full circle back to publishing literary criticism by reviving our once thriving Penn State Series in Romance Literatures and broadening it out into a Romance Studies series of monographs that will be “born digital.” Overseen by a joint faculty committee from the departments of French and Spanish, Italian, and Portuguese, supplemented by a broader advisory board of scholars from other universities, this series will publish original works of scholarship as well as translations online, making them available for use via the Internet to the entire world at no charge as PDFs with full-text search capability and with permanent URLs guaranteed by the libraries, which will be responsible for metadata, technical support, and archiving. The model we are following was initiated some years ago by the National Academies Press (NAP), which mounted all its titles online for free browsing, but with limitations (such as page-by-page downloading and low resolution for print) that provided people who wanted to do more than browse with incentives to purchase books printed “on demand.” Its success with NAP's titles, not cutting into book sales but even to some extent increasing them, is encouraging enough that it seems worth testing for books in the humanities and social sciences also. So anyone who visits our site and browses any title in our Romance Studies series will be able to click on a “buy” button and receive a copy printed by Lightning Source and shipped directly to the customer. Our assumption is that these POD sales will not be any lower than the sales that books in the previous series had reached on average before we suspended it and, with costs for this mainly online venture likely to be somewhat lower than for the traditional print model, a few hundred copies sold in this manner should suffice to sustain the program, perhaps with some modest level of subsidy as we had earlier been receiving for many books in the series from the Spanish Ministry of Culture. Yes, we will in the near term, at least, need to print out a small number of copies, probably around 50, to give the author at the outset and to send to review media, but we can avoid sending out the usual “exam” copies because every book will be readily accessible for browsing online and there will be no returns to deal with since the books will be sold as nonreturnable (as all of our titles in Lightning Source's system are now). For authors, there will be a tremendous advantage in having their work free for browsing anywhere in the world (which is what we mean here by “open access”), which may lead to many more citations and much wider use overall, while their promotion-and-tenure committees can still be presented with a physical book as tangible proof of their scholarly achievement. Ancillary materials can even be provided on the web site connected with their book, such as the text of the original work from which a translation was made, for instance, or color illustrations that would be too expensive to include in a regular print or POD edition. In this sense, opportunities remain to publish “hybrid” books, although they fall short of being quite the innovative e-books that Robert Darnton envisaged as using the web to its full publishing potential, but which would be too costly to produce under this business model. Still, it strikes us at the Press as a chance to rescue some of the “endangered species” of scholarship like literary criticism and to experiment with “open access” as a model for the future. Ironically, the “endangered species” under this model will leap from the one extreme of having only a few hundred copies of a book available worldwide in traditional print form to the other extreme of being immediately accessible to every student, teacher, and general reader in the world electronically, with the option to buy a POD copy. Is this a fate that one should wish for political theory? Does it take disaster to produce salvation? Stay tuned, for this story is only now getting under way and its conclusion has not yet been written.