February 15, 1997

The original SUL/AIR proposal to examine the impact of electronic publishing on the scholarly community focused on the users of electronic journals. The population for a series of empirical surveys was to be the subscribers to the Journal of Biological Chemistry (JBC), published by the American Society for Biochemistry and Molecular Biology (ASBMB) and the Libraries' HighWire Press. The journal is one of the most prestigious in the field and the Press's implementation of JBC's on-line version is regarded as a trail-blazing example of electronic publishing. Equally important in the choice of JBC for the case study was the fact that a full-text version has been on-line since May 1995. Hence it was reasonable to expect an informed reaction from the user community.

In the course of the initial investigations, however, it became apparent that users were only one of several identifiable groups whose decisions and reactions will be important in shaping the future of on-line publishing. For example, as might be expected, publishers play an important role in determining how rapidly and in what form electronic publishing evolves. Because publishers span a wide range of groups from scientific societies to commercial publishers, there is an equally wide range of discernable strategies as they strive for pricing models that will best serve their long-term interests. Although it is difficult to generalize, to date one effect is clearly visible. The overall impact of electronic publishing has been to provide publishers with an opportunity to increase revenues from the serials they publish. Because of the proprietary nature of publisher's costs, however, it is impossible to say if additional funds have been added to their bottom line.

It also became clear in the course of the investigation that libraries and librarians will play a powerful role in determining the rapidity and scope of e-journal adoption. Admittedly, the librarian's job description differs significantly for one organization to another. In small departmental or company libraries, for example, their decisions about what journals should be purchased and the type of display that should be provided are carried out in close consultation with researchers. However, as is evident from focus groups, a survey of institutional subscribers, and various mailing lists, librarians in the larger research libraries are crucially involved in negotiating publisher terms on a wide variety of issues, including costs, access, and what constitutes "fair use." As a generalization, the combination of the desire to control costs, a somewhat reluctant user community, and a well developed sense of the need for a reliable system of archiving, is making librarians cautious in purchasing e-journal subscriptions.

The report takes up each of the groups: users, publishers, and institutional subscribers (librarians) in turn and examines the issues associated with electronic publishing as these have been expressed in surveys, focus groups, and lists. Based on these empirical findings, the final section speculates on the most likely path of events over the next few years. It is a conservative prognosis, one that sees the emergence of electronic publishing as an evolutionary technology that will blossom at the same time as the dominant print technology remains in place.

The trend of 1996 in which large numbers of electronic journals appeared on the market is likely to continue unabated until virtually every journal in the "hard" sciences has an on-line version. But user surveys suggest that the impact that these journals will have relative to their print versions will be quite modest, a view that is consistent with the conclusions in the e-journal literature.

The empirical portion of this study focuses on an in-depth investigation of the attitudes and reactions of the readership of the Journal of Biological Chemistry. The full text of the journal went on line in May 1995. Hence it has one of the longest, if not the longest, on-line publication record of any major scientific journal.

The initial survey of JBC readers was designed in collaboration with an Associate Editor of JBC, Robert Simoni. It was administered to a sample of the JBC Editorial Board at its annual meeting held in conjunction with the meetings of the American Society for Biochemistry and Molecular Biology. The questionnaire, completed by 135 senior scientists, is reproduced in full with complete tabulations, and is appended as Appendix A.

The second round of the Editorial Board was carried out using approximately the same questionnaire, this time administered via e-mail. It was sent to the remainder of the Editorial Board who had valid e-mail addresses. There were 250 responses from a list of over 400 members for a 60 percent response rate. The results of this survey are reported in Appendix B.

Subsequently, a third round of the user survey questionnaire, reported in Appendix C, was administered to some 150 individual subscribers of the on-line version of JBC. Some 75 scientists responded for a 50 percent response rate.

A fourth and final round was administered to a sample of 500 randomly selected authors who published in the JBC in 1996 (Appendix D). The sample differed significantly from the Editorial Board sample in that, among other things, a much larger percentage of the authors was from foreign countries. This had the effect of (a) raising the interest of many researchers in the on-line version because they receive it more quickly, and (b) increasing the number of subscriber complaints about access problems.

The questions in the Author Survey also differed from those asked previously in that they concentrated more explicitly on the print versus on-line issue. Not only were readers asked about the advantages and disadvantages of each, but they were also asked specifically about their reactions to suggestions emanating from the JBC leadership that the print version would be discontinued in the future.

The paragraphs below summarize the major results that were common to all of the tabulations. Side commentaries suggest alternative conclusions where the results differed among the surveys.

1. The scientists who have access to JBC Online were, with rare exceptions, extremely complementary about the quality and style of its presentation. Especially in the Author Survey where a question was asked specifically about possible improvements, many answered that they could think of little that could be done better, given the limitations of the current technology. Several reported that JBC Online was the best electronic journal in their field and that it has had a trail-blazing effect, creating standards that other journals are now trying to emulate.

2. A substantial number of researchers have gained at least some experience with JBC Online although only a limited number use it exclusively, even when they have purchased the individual on-line subscription. The answers to usage question are reproduced in the table below.

The broadest generalization from these results is that approximately half of the researchers contacted used print only and the other half used both. However, there are several sources of bias in the numbers that make them difficult to use for projections about attitudes toward the on-line version of the journal.

1. Of the total of 3,700 institutional subscriptions, only 9 percent are subscribers to JBC Online. Yet nearly half of the respondents have some experience with the on-line format. This suggests that surveys directed at "readers" of JBC (as opposed to the Editorial Board and Authors) about such issues as print versus on-line, would have yielded more conservative conclusions concerning the extent that users are familiar with the on-line version than the Editorial Board and Author numbers indicate. Both of the aforementioned groups are likely to come from major research universities that have the resources to purchase both versions.
2. A bias in the opposite direction, however, that undercuts the "percent of total subscriptions" as a measure of user acceptance, arises from the lack of a one-to-one correlation between users and institutional subscriptions. Major research universities are likely to have only one on-line subscription but a large number of faculty and graduates students, most of whom use the on-line version. If the total subscription numbers were used as a basis for estimating the importance of on-line use, it would significantly understate the impact of the on-line version.
3. A further complication in interpretation arises because the surveys asked users about their experience during a time when the on-line version was free. It is conceivable that there were many scientists who used the free on-line version at that time, but who do not now use it because their institutions, for reasons of cost, do not subscribe to both versions.
4. It is also possible that many scientists use the portion of JBC Online that continues to be available to everyone (tables of contents, abstracts) and count that as a use of "both," even though their institution does not subscribe to the full-text version.

Perhaps the best that can be said at this point is that scientists appear to regard the print and on-line versions as complements rather than as substitutes. When asked about the advantages of JBC Online, they replied: rapidity of searching, convenience, timeliness, and efficiency of storage. When asked a similar question about the advantages of print, they answered: ease of reading, quality of figures, serendipity of discovery while browsing, independence from computer systems, and portability.

The fact that large numbers of scientists use both the print and the on-line versions raises a perplexing policy issue for JBC management. Proponents of moving to an on-line only version of JBC have suggested that the current responses of scientists reflect a period of transition. In this view, greater computer literacy and improvements in the technology of electronic delivery for both on-screen and hardcopy will make the bound copy superfluous.

But the persistent differences cited by respondents about why they like the printed version, including the repeated references to the role of serendipity while browsing the printed version, lend creditability to the argument that the two are different ways of obtaining information, each with its own strengths and weaknesses. A number of scientists indicated that they believe that all journals are going to be on-line only in the future. But the majority are quite adamant that they envisage this future to mean "distant future." In the survey of authors, a near majority of a biased sample (46 percent) selected the option: "continue for the foreseeable future" as their preference on the issue.

3. Feelings among scientists surveyed regarding the print versus on-line question are quite intense. With an innovation still in its infancy, this is to be expected. Those who have computer facilities and the skills that make it easy for them to access the on-line version are positive about the experience; they are solidly in favor of having the on-line version, even if they do not favor the on-line only option. But for the group of scientists--one presumes the older generation--who did not grow up with computers or who do not have access to the state-of-the-art technology required to read on-line materials comfortably, moving to an on-line only version would make their access to the JBC much more difficult.

In the final (author) survey, scientists were asked whether a decision to move to an on-line only version would adversely affect their decision to publish in the journal. Approximately 35 percent replied that it would. This result could be expected from the fact that, as the table shows, a substantial number of authors do not use to the on-line version at all, many because they do not have access.

There was a countervailing weight on the other side of the issue, however, in the 15 percent of authors who reported that they were "somewhat" positively influenced in their decision to submit their 1996 articles to the JBC.

4. The professional reading of the group of scientists that subscribe to JBC is highly focused. Indeed, one could construct a "knowledge environment" from this group that would encompass most of what they read by focusing on a half-dozen journals. Most prominent among the additional journals would be Science and Nature. Other major journals that would round out the remainder of the half-dozen would be Journal of Molecular Biology, Cell, and PNAS.

Publishers often do not have a business model in mind when they begin to move their content on-line. Part of their on-line migration process is to gather business model information and print-subscription pattern data. A key concern is that on-line institutional subscriptions will cannibalize individual subscription and institutional subscription revenues. Some publishers are attempting to protect or even expand their personal subscription base through the on-line market by offering "personal services". Personal services include e-mail alerts of new on-line content, prepublication of articles, personal subject profile searches, etc.

From the publishers' perspective the barriers to librarians' acceptance of the on-line editions (including user acceptance, archiving, ownership, and access) create the opportunity for a print-linked transition pricing model. These institutional barriers will surely erode over time, but they provide publishers a cushion of 3-5 years in which to develop new business models that rely less heavily on revenues from print.

Publishers are also concerned that pay-per-view services, which allow non-subscribers to obtain full text, will erode the subscriber base, as marginal readers decide to pay only for what they need "just in time" rather than have a full subscription "just in case". Similar concerns surround the interaction of pay-per-view with reprint income.

Currently, it is common practice for publishers to provide tables of contents and abstracts for free to all network users. However, there is an increasing fear in some quarters that this policy is eliminating a good deal of the value delivered to subscribers since an abstract is often quite sufficient to communicate the key results of a paper. As a market penetration strategy, providing a certain amount of "free" information may be desirable. It may not, however, be a practical strategy once the use of electronic journals has become more widely accepted.

The following paragraphs provide a description of common subscription models/pricing (how things are priced) and access models/controls (how access is controlled) for individual and institutional subscribers. As yet, there is insufficient evidence on the pay-per-view model to do more than speculate on its acceptability.

Presented below are six illustrative subscription models; there are infinite variations on these themes.

In cases where print and on-line contents are similar, librarians do not like "forced bundled" models in which the on-line version is bundled with a required print copy. For example, the University of California on-line collection development prohibits purchase under this model. Librarians perceive that "we're being forced to pay for the same content twice." Instead of making the cost of the on-line version explicit, some publishers are raising print prices by 5% to 15% and allowing "free" on-line access (e.g., Journal of Neuroscience, Journal of Applied Physiology). Biotechnology Letters, published by Chapman Hall (6 issues in 1996), charges $413 for print and on-line and $344 for Print Only, a surcharge of about 17%.

Some publishers are delivering different content in their print and on-line versions. For example, Pediatrics On-line is comprised of abstracts from the print journal plus 6 to 10 additional full-text electronic-only articles. To access all the content from this title, access to both the paper and the electronic versions is required. (Pediatrics will raise institutional prices by approximately 5% to cover the electronic content costs.)

The reaction to this strategy depends upon the additional cost of the on-line version. Librarians generally "flag" journals with price increases equal to or more than 10% for possible cancellation.

The JBC Online business model offers a choice: print ($1,400), on-line ($1,100), or both ($2,500). In principle, this pricing model addresses the librarian's reluctance to pay twice for the same content. If, however, they feel they must choose both print and on-line versions, as many feel they must, the combined cost is substantial.

The ASBMB sees this pricing policy as a way of accelerating the transition from print to on-line by creating a financial incentive for librarians to adopt an on-line only strategy. To date, the evidence suggests that librarians who don't have the resources to pay for both print and on-line editions are choosing print. Canceling print and getting only the on-line edition seems too risky given concerns about archiving and access.

Subscriptions to IEE On-line Journals are priced the same as the equivalent print versions. This covers unlimited on-line searching, display, and local printing by users at a single site. The price for a combined print and on-line subscription to one or more IEE Proceedings titles or Electronics Letters is 1.5 times the print price. The more titles an institution purchases, the more money they save:

In general, librarians respond favorably to this model because they have choice of formats and there are economies of scale. However, on-line versions that have surcharges of over 15 percent are not usually purchased by librarians.

American Chemical Society's journal price depends on whether a print and/or an on-line subscription is purchased. For example:

Although some librarians may like the flexibility offered by this model, many will be overwhelmed by the complexity of their campus networks and decisions about the level of demand from particular groups.

Academic Press will not sell individual, by-title subscriptions to libraries unless they are part of a consortium and buy a majority of the AP list. Their AP IDEAL package includes 175 journals. The full text is in Adobe Acrobat format. There are two exceptions to the AP consortium restriction: J. Molecular Biology On-line and Genomics On-line were available free to institutional print subscribers through 12/31/96. There is no pricing information available on the web, but the "librarian grapevine" puts the on-line surcharge for these titles at least 5% above the printed cost. This approach makes sense for Academic Press. They have a large suite of journals, most of them of secondary importance, most of them not likely to be purchased individually.

Librarian reaction to this model is mixed. Some librarians enjoy having the consortia make decisions and arrange access. Others have decided that these titles are not important enough to "buy twice," a reaction aimed less at the business model and more at the specific journals that are offered.

MathSciNet is a web database based on the data in Mathematical Reviews and Current Mathematical Publications. The subscription price for Mathematical Reviews (MR) and MathSci products separates the cost of developing and maintaining the MR bibliographic database of editorial content (from which all products are derived) from the cost of the individual product deliveries. Subscribing sites contribute to the development and maintenance of the database by paying an annual Data Access Fee (DAF). Sites and individuals associated with those sites can then subscribe to MR products in a variety of formats by paying a Product Delivery Fee. So, for example, an institution might pay a $800 DAF for a title, then an additional $300 for each paper copy, and an additional $400 for an on-line site license, and an additional $100 for each CD-ROM. Librarian response to this model has been generally positive.

IP Based: All computers at a subscribing site may be authorized to access a journal. The publisher defines Îa site'. For example, the ASBMB defines a site as "an organizational unit within one city." Organizations located in more than one city are each a different site. (This definition is to ensure that large state university systems must purchase one license for each campus.)

Network Class Based: Only authorized networks at a subscribing site may access a journal. The American Chemical Society defines a site as: one geographical campus or building; corporate sites are limited to one company per building or campus. If a building or corporate campus has two separate companies, including one which is a subsidiary of the other, these are viewed as two sites. Fees are based on the number and kind of network configuration within a site. A print subscription authorizes (i.e., includes) access for one Class C license. (A Class C network might include a workgroup in a single building; a Class B license will usually include a small institution.) In the example above, an additional Class B license for a non-member is $1,955.

Site licenses in some form are preferred by virtually all librarians because they reduce administrative costs to a minimum. Potentially, there are significant economies of scale involved that has prompted some publishers to calibrate license fees to reflect the size of the institution.

This option is well established for traditional database access, but is unclear as a journal subscription model. For example, the 1996 MathSci Disc Network Option adds the following surcharges to the MathSci Disc first copy price. Prices are for each disc networked. Per-user pricing is as follows:

This model holds both appeal and potential risks for publishers. A perceived advantage is that they can provide institutional-wide access while limiting use. Fees can be set proportional to use. One risk is that this scheme allows librarians to be very price sensitive. Librarians may choose a subscription level based on what they are willing to pay, not necessarily on service levels. Another risk is the number of simultaneous access ports needed, for even the most popular science journal is not likely to be very high. According to participants at librarian focus group sessions, very large research institutions are unlikely to purchase more than 3 simultaneous "seats" for the most important, popular science titles.

Springer-Verlag is offering institutions "free" electronic editions in addition to their print subscriptions. Institutions sign an on-line license agreement for up to 25 concurrent users. Licenses to store journal data files and to disseminate them within the networks of corporations or cooperating libraries will be available upon request.

Commercial librarians favor this model because they have fewer users of any particular title. Academic librarians like it because it allows them to be "price sensitive".

In 1996, individual readers at institutions with print subscriptions to Nucleic Acids Research (NAR) could register free of charge for full access to the NAR On-line. However, every person needed a distinct username and password. They had to obtain their institutional subscriber number from the department (e.g. the library) where the printed copy of NAR is mailed. This model is discouraged by librarians because of the need to handle usernames and passwords.

This model authorizes particular situated machines within an institution with the library usually suggested as the proper location. Discussions with librarian focus groups show this to be an unacceptable model.

Models for individual subscribers follow many of the institutional models. In addition, a recent trend is to segment and differentiate the individual/member from institutional "users" by offering "personal services". There are at least six market segments:

1. Very limited free content available to entire web, example: Science Magazine
2. Free full text to entire web, example: Journal of Clinical Investigation.
3. Somewhat limited content available via web with questionnaire registration, e.g., Science Magazine
4. Full text and enhanced services (e.g., email notification) available as a premium paid service. Designed to attract and keep personal subscribers who also have on-line journal access via an institutional subscription, e.g., Journal of Biological Chemistry.
5. Pay-per-view . Users enter their credit card numbers to pay for content. Pay-per-view options are available for article viewing, downloading and document delivery. Sales can be per article, or a number of articles purchased at a discount off the single-article price.
6. Full text available to a particular category of people:

1. Members/individuals who subscribe to print receive full text on-line for free
2. Members get full text on-line for free as part of their membership dues
3. Individuals/members who buy print get on-line full text for small surcharge
4. Members have option of buying on-line only

For example, American Chemical Society's journal price varies with whether a print and/or on-line subscription is purchased, and if the buyer is an ACS member. The on-line edition is cheaper when purchased with a print subscription. For example:

Access using an IP address requires only that the user register the address of a single machine. No other identification is required. In systems requiring a user name and password, access is granted when payment is confirmed. Users are then assigned an authorization number (which may be their membership or subscriber number). This number allows them to enter the system and choose a username and password. Individuals are able to change and update this information on-line. If they forget their password, they may enter either their username or authorization number and the system will assign them a new password. They then can change the machine-assigned password to one they choose.

Although the sample of publishers reviewed for this report is quite small÷next year there will be many more÷the evidence suggests two propositions:

1. Journal publishers are using the emergence of electronic formats as a device for collecting market information while at the same time they are enhancing revenues. When evaluated in terms of impact per page, many of the high-impact journals have been under-priced relative to their less well-known competitors. However, they have been constrained in their ability to increase prices by their reluctance to alienate the community of which they are a part. The on-line version, especially when it contains enhancements that go beyond the print version, has enabled them to leverage the substantial investments they have in their publishing activities.

The unknown in the equation is what effect the increase in revenues is having on the bottom line. Data on costs are closely guarded proprietary information. Hence it is impossible to say whether, as some have suggested, the emergence of on-line versions is a public service and that increased revenues only cover development costs or, as others suspect, increased gross revenues are also yielding increased net revenues.

2. The overall increase in serial prices resulting from the additional charges for on-line displays will, in the face of stagnant library budgets, put substantial pressures on second-tier journals that are often part of the stable of large commercial publishers. So long as archiving problems remain unresolved and the resistance of users to abandoning print entirely continues, few institutions will give up their print titles. Many, however, will decide to make modest investments in the on-line versions of major journals. To make room for these expenditures, they will reduce the number of second-tier print titles and specialized monographs.

Libraries and librarians link the readers of today with the authors of today and yesterday. The electronic environment has placed additional stresses on this traditional role by creating vagueness and uncertainty in a number of key areas. These include:

1. The choices (and costs) associated with maintaining a constantly changing technological infrastructure. The computing environment directly determines if readers can successfully use electronic materials. More often than not, librarians are disconnected from the responsibility and budgets for key pieces of this infrastructure. Few librarians have control over campus networks, desktop computers, and printers, etc.; most find their current infrastructure is inadequate to support the increased demand generated by electronic journals.

2. A value system that places a premium on enfranchising all user groups. Librarians are staunch protectors of all classes of readers and collection decisions are intended to broaden access to information. Substantial gaps have arisen among users with differential access to technology and the have-nots find a sympathetic when they complain to librarians that they don't have the equipment or software to access electronic materials. Librarians are likely, in the face of an inability to solve the infrastructure problem, to be cautious in making trade-offs that might disenfranchise some readers.

3. Stagnant library budgets that require giving up old, often popular programs to undertake new initiatives. The limited size of the library pie and the expenditures required to move to more innovative solutions to scholarly communications have meant that future expansion must come through painful retrenchment.

4. A wide variety of publisher business models that has increased the amount of time that must be devoted to vendor relationships. As the previous section on publisher strategies suggested, many publishers are, with their initial proposals, simply testing the market for their on-line products. Understanding and responding to the range of proposals currently on the table requires substantial time taken from other duties.

5. Newly emerging issues of intellectual property that create uncertainty about the limits of well established print guidelines to fair use of library materials.

Data from several sources have been used to gain an understanding of how librarians are responding to the issues raised above. The primary source has been the results of a survey of JBC's institutional subscribers. The libraries of institutions and organizations account for a majority of scientific journal subscriptions and JBC is no exception. There are approximately 3,300 institutional subscriptions compared to 200 individual subscriptions. It is reasonable to assume, therefore, that the perceptions of professional librarians about electronic journals will be an important determinant of whether the organization subscribes to the JBC or not.

The results of the survey have been supplemented by the findings of focus groups conducted in the Stanford area by Science magazine. The groups represented librarians drawn from both public and private research organizations. A number of mailing lists devoted to electronic publishing have also been sampled. These include ARL-JOURNAL, DIGLIB, and the newly established LIBLICENSE.

The questionnaire, the results of which are discussed below, was sent to the e-mail addresses of approximately 300 institutional subscribers. Because there was no indication on the addresses of the status of the individual originating the subscription, recipients were asked to forward the questionnaire to the appropriate person responsible for making subscription decisions.

Approximately 60 recipients responded for a 20 percent response rate. This is considerably lower than the response rate of scientists in the Editorial Board and individual subscriber surveys. The lower rate is due in part to the large number of foreign institutions that subscribe to JBC. They are substantially under-represented in the results. It may also be due to the fact that the questionnaire was sent to the address of the person who initiated the subscription and not the individual(s) responsible for decisions regarding subscriptions.

The results of the survey suggest several general propositions:

1. The distribution of answers to the question of organizational e-journal subscriptions was distinctly bimodal. Seventy-five percent of the respondents replied they had "less than 5" institutional subscriptions. Fifteen percent reported more than 20. The remaining 15 percent were divided among the intervening alternatives.

This finding underscores the variation across institutional subscribers that permeates all answers to the librarian questionnaire. As the subsequent evidence suggests, the number of electronic subscriptions, the decision-making role of librarians, and the relationship between librarians, users, and publishers are very much a function of the size of the organization. In smaller groups, including corporations, the role of the librarian resembles a clerical function--someone who orders materials as directed and catalogs them when they arrive. In the larger research libraries, the distance between users and the librarians is greater and the latter play a more significant role in determining policies regarding on-line subscriptions.

The differences in the roles that librarians play is also evident in the answers to the direct question of who makes decisions about the purchase of electronic subscriptions. Over 40 percent of the librarians answered that the cognizant librarian made the decision. Although no formal correlation analysis was done, examination of the questionnaires indicates that these respondents have e-mail addresses that suggest they represent the larger research universities. The same is true of the 20 percent who said that a library-wide policy committee made the decision. The 30 percent who reported that some sort of joint committee of scientists and librarians made subscription decisions were located in departments or, frequently, in what appear to be commercial research labs.

These results are not to suggest that users play no role in the purchasing of journal subscriptions of major research centers. However, the interactions are likely to be more indirect and subscription decisions are heavily filtered by the librarian's institutional concerns about costs, archiving, duplication of information, etc.

2. Librarians are, and will continue to be, the major gatekeepers to electronic publishing where issues of cost are concerned. In a few cases, survey respondents reported cost sharing arrangements with laboratories and departments. But for the most part, it is librarians who must balance the increased cost of adding the electronic version of a journal to their holdings against competing needs to maintain the breadth of their print serials and monograph collection.

The additional costs of e-journals, even the modest costs of bundled versions, come on top of a decade-long "normal" increase in serials prices of roughly 15 percent per annum. This long period of rising prices has meant that most libraries have had to go through several rounds of journal culling already. The survey indicates that most institutions have developed fairly systematic approaches to title reductions, but each such exercise is difficult because it requires often imprecise evaluations of journal use and, ultimately, a reduction in holdings.

The survey also indicates that librarians are clear that adding an e-journal version of a print title means that the same evaluation procedures will be required, but this time applied to the value-added by the e-version. Prior to 1996, this was not a serious problem because so few electronic formats existed. In the future, the issue of assessing the value-added by the electronic version against maintaining the current print collection will become more serious. It will be further complicated if publishers enhance e-versions of journals and use the flexibility of on-line presentations to develop a more dynamic approach to publishing.

The generally accepted notion that, in periods of rapid change, old and new technologies may exist side-by-side is of no consolation in the presence of materials budget constraints. Looked at from a model of bureaucratic risk taking in which penalties for budget shortfalls may be significant, the result is likely to be a conservative approach to change. The prediction of such a model would be that librarians will move slowly and cautiously to substitute e-journals for print if it means substantial further declines in their already reduced serial and specialized monograph holdings. This is especially true of units where the library itself must provide the infrastructure for viewing the electronic version of the journal on-line.

3. Although comments in the survey indicate that librarians consider cost to be extremely important, they are also concerned about the entire issue of archiving and ownership. Opinions about the best way to proceed differ, as the ARL-JOURNAL mailing list devoted to this thread makes clear. The lack of a cost-effective solution to the problem that everyone: users, libraries, and publishers, can live with is likely to create a significant incentive to retain print versions, at least in the short to medium run. Again the model of bureaucratic risk is helpful. Reliance on a system of archiving that is based on relatively unproven technical and institutional arrangements can lead to serious criticism whose consequences to librarians are greater than the benefits of providing on-line access.

4. At the same time that issues of cost and archiving suggest a cautious approach by librarians to the acquisition of e-journals, the survey and the mailing lists indicate that the library community is intensely engaged in a practical way in the changes that are taking place. Nowhere is this more evident than in the debate of issues of intellectual property and fair use that the electronic journals create. As the name suggests, an entire mailing list, LIBLICENSE, is devoted to topics that, for the most part, revolve around the legal and financial relationships between libraries, e-journal publishers, and vendors. The comments in the list as well as those made by librarians in the institutional subscribers survey indicate a high degree of sophistication among librarians regarding these issues. Again, there is no unanimity about the pros and cons of various publisher business models or what constitutes fair use. There are, however, increasing efforts to share information that would lead to "standards" regarding such issues as fair use, inter-library loans, and archiving procedures that would be fair to all parties.

5. Users create the demand for electronic publications; publishers create the supply. But except in a few cases, librarians create the "market" that brings the two together. This has been their traditional role and, with the current state of technology and division of labor, that role is likely continue. What has sometimes gone unappreciated by the focus on users and publishers is that this critical role is carried out within a set of budget and institutional constraints that have a profound influence on the market's outcome.

Although the basic empirical work of this study was confined to the readership of a single journal, i.e., the Journal of Biological Chemistry, its prestige as well as its early entry into the on-line market made it a good case study candidate. JBC's appearance and functionality have been the defining standard, not only for journals published by HighWire Press, but for the on-line versions of other science, medical, and technology publishers as well. There is good reason to suppose that the insights reported here will be relevant for the larger market as it emerges in the next few years.

If the past year is a guide, in the near future every STM publisher will be bringing out on-line versions of their journals. Competition for the best authors demands that this form of information dissemination be part of the package that makes up a publisher's or society's services. Most publishers have and will use the opportunity to leverage their print versions into on-line versions on which additional reader charges can be levied. The added costs of on-line versions of major journals will force libraries, already under severe budgetary pressure, to further reduce secondary serial titles and to curtail the acquisition of specialized monographs.

What remains to be seen is how rapidly the research community will be willing to dispense with print in a way that will give libraries leverage in negotiating with publishers for reduced costs. The archival problem looms large in this decision and the absence of trustworthy technical and institutional solutions is likely to create a reluctance by libraries to abandon print even when the research community has moved beyond the need for hardcopy.