This precis of "Towards Electronic Journals" (Tenopir & King 2000) focuses mostly on scientists' perspective as authors and readers, how changes over the years by publishers and librarians have affected scientists, and what they should expect from electronic journal and digital journal article databases. We describe some myths concerning scholarly journals and attempt to assess the future in a realistic manner. Most of our primary data involves U.S. scientists, libraries and publishers, but much of the secondary data is from a European perspective, which shows few differences.
PSYCOLOQUY CALL FOR BOOK REVIEWERS: Below is the Precis of "Towards Electronic Journals" by Carol Tenopir and Donald W. King (900 lines). This book has been selected for multiple review in Psycoloquy. If you wish to submit a formal book review please write to email@example.com indicating what expertise you would bring to bear on reviewing the book if you were selected to review it.
(If you have never reviewed for PSYCOLOQUY or Behavioral & Brain Sciences before, it would be helpful if you could also append a copy of your CV to your inquiry.) If you are selected as one of the reviewers and do not have a copy of the book, you will be sent a copy of the book directly by the publisher (please let us know if you have a copy already). Reviews may also be submitted without invitation, but all reviews will be refereed. The author will reply to all accepted reviews.
FULL PSYCOLOQUY BOOK REVIEW INSTRUCTIONS AT:
Psycoloquy reviews are of the book not the Precis. Length should be about 200 lines [c. 1800 words], with a short abstract (about 50 words), an indexable title, and reviewer's full name and institutional address, email and Home Page URL. All references that are electronically accessible should also have URLs.
AUTHORS' RATIONALE FOR SOLICITING MULTIPLE REVIEW: We would like scientists as authors, readers, editors, referees and observers of the coming electronic age to review the book through their personal experiences and knowledge, which they think confirm, reinforce, or refute our observations. We would also appreciate comments on our interpretation of results. We look at the book as a stepping-stone in our further study of electronic journals. Input from scientists is particularly desired for our future study.
OVERVIEW OF CONTENTS: "Towards Electronic Journals" (Tenopir & King 2000) is addressed to four audiences: scientists as authors and readers; journal publishers; librarians and other intermediaries; and organizational funders of scientists and libraries. An attempt was made: (1) to describe the communication practices of scientists, librarians, and publishers; (2) to establish their goals, motives, and incentives for the way in which they do things; and (3) to determine the cost and other economic aspects of their involvement. In particular, we felt it important for each journal system participant to gain a better understanding and appreciation of the contributions made by all participants and to enable them to make more informed decisions about electronic journals in the future. To achieve these objectives we partitioned the book into five parts, in addition to an introduction. A background part provides a summary of the quantitative results, a brief history of scientific scholarly journals including early electronic publishing, a framework for describing scholarly journals as a system embedded in larger communication and science systems, and a description of our data collection methods. Data include results from 13,591 readership survey responses from scientists (1977 to 1998); more than 100 cost studies of library services, publishing, and scientists' authorship and information seeking; a study of the characteristics of a sample of 715 scholarly journals tracked from 1960 to 1995; and review of more than 800 relevant publications. The next three parts address the principal participants: (1) scientists, including their general communication activities and journal authorship, readership and information-seeking patterns; (2) libraries, including general library use and journal-related services use and economics; and (3) publishers, including journal publishing costs, pricing, and financial considerations. The last part covers electronic publishing details and aspects appropriate to each of the journal system participants.
1. Science is undergoing some fundamental changes. Much of science is experiencing greater specialization, while, on the other hand, some parts of big science are becoming even bigger. A great deal of research is becoming more multidisciplinary, often involving a range of in-depth specialties and requiring collaboration among universities, government and industry that extends across national borders. Science education is also becoming multidisciplinary and collaboratory, as more and more faculties teach across disciplines, departments, and universities. Many companies form project or product teams from different disciplines to follow products from discovery to the marketplace. Science research and education rely more on innovative instrumentation and communication technologies, which affect communication patterns among individual scientists and communities of scientists in fundamental ways. The Internet and other advanced technologies have opened the flood-gates to an enormous flow of data, images, and text.
2. Learning is fundamental to science, whether through research and discovery or through education and other forms of lifelong learning, and communication is at the heart of learning. Garvey (1979), a psychologist at The Johns Hopkins University, summed up nearly two decades of scientific communication research by concluding that "communication is the essence of science". Fourteen independent studies conducted from 1958 to 1998 all observed that scientists spend a large, and perhaps increasing, proportion of their time communicating, thus serving as an indicator of its increased importance to them. Our most recent studies place this proportion in the range of 50 to 60 percent of scientists' time spent communicating (on average). There is some evidence that scientists in industry spend about 150 more hours per year performing their work than 15 years ago, largely as a result of electronic messaging and informal meetings. Thus, the capacity of scientists' time may be reaching a limit that must be dealt with or at least seriously considered through communication systems in the future. Science and scientific communication are clearly intertwined and are affected by their respective changing environments.
3. Traditional scientific communication patterns have evolved into a multitude of channels, including data and image transmissions, informal discussions and electronic messages, laboratory notes and technical reports, conference presentations and proceedings, journal articles, patents and books, to name a few. Within each channel are several potential distribution means. For example, scholarly journals are communicated through personal and institutional subscriptions and separate copies of articles through preprints, reprints, interlibrary loan, document delivery, and copies provided by authors and colleagues. Each means of distribution can involve a variety of media. For example, journal subscriptions are available in paper, CD-ROM, online and microform; separate copies are available in offprints, photocopies, facsimile, and online. Together all these channels and the means used for their distribution form an extensive and complex pattern of information flow.
4. The American Psychological Association (APA) performed a series of highly regarded and oft-cited communication studies for the National Science Foundation (NSF) in the 1960s [American Psychological Association 19631968], and later in the 1970s, involving a variety of scientific disciplines in addition to psychology. Among extensive observations, these studies carefully described the flow and timing of the communication of research results through the various channels, including the extent of redundancy of information communicated through different channels. These studies revealed that a particular research result is often communicated through a variety of channels over time, sometimes modified as it passes from channel to channel as a result of feedback from other scientists and from further research refinement. Some results of scientific studies were also repeated in as many as eight scholarly journals, albeit with different emphases or with different audiences in mind.
5. As early as the 1960s, these studies led many to believe that electronic technologies could minimize redundancies and produce more efficient communication capabilities. It was thought that once text was initially digitally captured in reports, conference proceedings, and/or journal articles, little further input would be necessary if the digital text resided in an accessible public "database". During the 1960s, some even predicted the imminent demise of scholarly journals. A great deal of NSF and other government-supported research was devoted to achieving these ends with little to show for it. However, technologies had not reached current capabilities and their use was not nearly as widespread as now. The concerns and hopes expressed in the 1960s and 1970s now appear possible, particularly through scholarly journals and an accessible public digital text database or databases.
6. The first two scientific scholarly journals were believed to have started at about the same time, in the mid1600s. One was Le Journal des Scavans, which was founded by M. de Sallo, a counselor of the French court of the Parliament. The second was Philosophical Transactions, a monthly journal of articles provided by the Royal Society of London to its members. The first issue contained sixteen pages consisting of nine articles, a dedication to the Society, a listing of books, and other correspondence. By the end of the 17th century there were about 30 to 90 scientific and medical journals published worldwide and this number rose to about 750 by the end of the 18th century. Currently there are thought to be about 80,000 to 100,000 scholarly journals published worldwide. While several communication researchers and science historians have tried to pin these numbers down, they have been unable to develop accurate and precise estimates due to definitional problems and an inability adequately to track journal births, deaths, and the "twigging" of journals into two or more titles.
7. The exact date of the founding of the first scientific journal in the U.S. is a matter of contention. Various reports of the first such journal are: one volume of Transactions produced by the Chemical Society (1813); American Mechanics Magazine (now Journal of the Franklin Institute), published in 1825; an agricultural science journal (1839); and Scientific American (1845). The growth of U.S. journals exceeded worldwide growth until about 1875, from which time it parallelled worldwide growth until the Great Depression in the 1930s. After that time the rate of increase appeared to be steady and linear on a logarithmic scale, until recently. We began tracking a sample of scientific scholarly journals (as a part of NSF contracts) from 1960 to 1995, with estimates of 2,816 such journals in 1960 and 6,771 in 1995.
8. While scientific journals have flaws, they have been said to be "the most successful and ubiquitous carriers of scientific information in the entire history of science" (Kessler 1967). Indeed, literally hundreds of studies have demonstrated their use, usefulness, and value. Many of these studies were performed in the 1960s and 1970s up to the time NSF stopped funding scientific information research. However, recent surveys of scientists in the U.S. confirm that early findings continue to this day. Scientists surveyed from 1993 to 1998 averaged 120 readings of scholarly articles per year. Three-quarters of readings by university scientists are for research purposes and over one-half of these readings are said to be essential to that activity. Two-fifths are for teaching purposes with most considered essential to that purpose. Nonacademic scientists read less, but information is extremely important to their work. In both settings, scientists whose work has been recognized through achievement awards read more on average than nonaward winners. A high proportion of readings are said to improve the quality of research and teaching, to help them do the work faster, and to save them time and money. We derived five indicators of productivity and these measures were all positively correlated with the amount of reading in each of six organizations in which observations were made.
9. Scientists, on average, spend well over 100 hours reading scholarly articles. They would not expend this scarce resource of their time if the information gained was not of value to them. The amount of scientific knowledge recorded in scientific journals doubles about every fifteen to seventeen years. Thus, at the time they graduate from college, scientists will have been exposed to only a fraction of the knowledge they will need during their careers, in fact, only about one-sixth of new knowledge created after their graduation. They must keep up-to-date or risk not achieving their full potential in research and teaching.
10. A number of visionaries in the 1930s and 1940s conceived of innovative ways of communicating, but many of these ideas were constrained by limitations to technology that no longer hold. These ideas were the genesis of a substantial amount of research performed in the U.S. in the 1960s and 1970s, largely under the auspices of the NSF and other government agencies. This research focused mostly on three areas: automated information retrieval systems; general scientific communication patterns and innovations; and the potential of electronic journals. In fact, by the end of the 1960s most of the components of a comprehensive electronic journal system were in place but were too costly, not widely used, and they badly needed standardized input and processing capabilities. When NSF funding for this research wound down (1978 to 1982), most of the initiative concerning electronic journals was carried on in Europe in the 1980s; particularly by the UK research community and a few commercial publishers. Some U.S. societies continued this research as well, but with little external funding.
11. One important aspect of the research from the 1960s onward dealt with the potential of a central database of articles. To achieve this end several experiments in the 1960s involved preprint and other forms of article distribution. Other researchers identified the need to complement (not replace) journal subscriptions with an inexpensive and efficient means of obtaining separate copies of articles. In fact, in the late 1970s, the U.S. Congress considered legislation to fund a central "National Periodicals Center", but the concept was defeated through heavy lobbying by some publishers and large libraries. Many researchers felt that a central digital database of articles would ultimately replace traditional journal subscriptions, minimize redundant research reporting among communication channels, and provide the means for innovative communication processes beyond those achieved through traditional channels.
12. By the early 1990s some scientists began to fill the void in electronic publishing research (assumed earlier through government funding) by building their own electronic journals (e.g. Harnad's Psycoloquy <http://www.cogsci.soton.ac.uk/psycoloquy/>) and digital databases (e.g., Ginsparg's preprint and archive database developed at the Los Alamos National Laboratory <http://ariv.org>). The Internet, World Wide Web, computer and software capabilities, digital input standards and codes (SGML, HTML), and widespread use of desktop computing all made comprehensive electronic journals and complementary digital article databases a reality. Aspects of the digital library initiatives also contributed to the enthusiasm of scientists, publishers, librarians, and other intermediary services. However, as often happens, the initial hyperbole exceeded reality but, on the other hand, the potential for innovation made possible by the technologies has not yet been realised.
13. A large proportion of scientific scholarly journals are now available in electronic media. Most are digital replicas of traditional journals, with the majority provided in both paper-based and electronic media and a few are exclusively electronic journals. Acceptance by authors and the number of readers have been disappointing to some, but such conclusions probably should not be, based on recent experience of scholarly journals. Trends in specialization of science and correspondingly smaller journal audiences have generally led to lower circulation of scholarly journals. Start up journals, whether paper-based or in electronic medium, tend to have very low circulation that increases much more slowly than in the past. Low circulation requires higher prices, discourages potential author interest, and leads to most reading coming from library-provided journals and articles. Electronic journals and digital databases should overcome these difficulties, but acceptance may continue slowly. Economic issues, particularly pricing, may be much more important in the future than acceptance of the new technologies.
14. Despite three and a half centuries of valued use, there has been a great deal of ambivalence and controversy over the years concerning the use, usefulness, and value of scientific scholarly journals. During the 1960s and beyond many have suggested that scholarly journals are no longer needed because they exist largely to fulfill academic tenure requirements, they only help feed "the information explosion", everyone is already aware of the information, they are simply not read, and they are slow to reach their audience. Furthermore, because of spiraling prices over the past 25 years they are no longer considered to be economically viable. Some of these criticisms of scholarly journals are entirely valid, but others are merely myths that should be dispelled. In order to address the potential of electronic media and what they can contribute to communication, one should carefully examine these perceived flaws to assess which are real and which imagined.
15. The amount of authorship of scholarly articles has remained stable over the years, although the number of authors per article has increased and a somewhat higher proportion of articles are written by university scientists than in the past. Authors, in surveys, report that they and coauthors together spend an average of about 80 to 100 hours preparing manuscripts which, when other resources are included, comes to approximately $6,000 per article published in 1998 (or about $6.70 per article reading). There are many reasons why scientists are willing to devote their valuable time to writing articles. Without a doubt, career advancement and tenure consideration are important to many authors, and some publish in order protect their ideas from theft or misuse. However, it is also true that altruistic motives such as contributing to knowledge and publishing for posterity are equally, if not more, important.
16. A great deal has been said about the proliferation of information, which in the past was frequently called "the information explosion". From 1960 to 1995 in the U.S., however, the growth of the journal literature has been highly correlated with the number of scientists, and this phenomenon appears to be true elsewhere as well. The number of articles published per scientist, while only about one article per nine or ten scientists, has remained relatively constant and pages published per scientist seems to have increased somewhat. The notion of an information explosion also fueled another criticism, which is that an average of only 10 to 15 percent of articles distributed through personal subscriptions are read. Studies over the past 40 years have been fairly consistent with this observation. Our surveys from 1993 to 1998 show that the average number of articles per journal is 123 and there is an average of 13.4 readings from personal subscriptions (or about 11% of articles distributed). However, there are many more readings from library subscriptions, typically over ten times as many. Furthermore, about 18 percent of readings are from separate copies of articles provided to scientists.
17. One criticism of scholarly journals is that most readers already know about the research through "invisible colleges", presentations at conferences, and other channels. While readership studies confirm that some scientists sometimes know about the research being reported, most readers do not when they first read articles. It is a myth that scholarly journals merely provide a means of communication among authors. In fact, particularly in the physical and life sciences, most reading is done by nonacademics working in industry, government, and national laboratories. That is, about 70 percent of all readings of scientific scholarly journals are done by nonacademics because most scientists work in nonacademic settings.
18. There are several bases for the statements that scholarly journals are not read much. One reason is that citations are considered by some to be the principal indicator of the extent to which articles are read. While several studies have demonstrated a positive correlation between article citation counts and observed use and reading, the citation counts represent neither total amount of reading nor the totality of readers. The single most influential source of statements concerning low readership of scientific articles and journals emanated from the APA studies done for NSF in the 1960s. Perhaps the most quoted results regarding readership of the articles are from a study by Garvey and Griffith (1963), in which they sent lists of article titles which were recently published to scientists and asked them to indicate which ones they had read. One survey result from among several disciplines, which continued to be quoted as late as the 1990s, is as follows:
"During the two months after publication, the audience for most articles is very small. About half of the research reports in core psychology journals are likely to be read (partially or entirely) by 19% or less of a random sample of psychologists. No research report is likely to be read during this period by more than 7% of a random sample of psychologists."
19. They reported that a typical article in the APA's journals was read by an average of only 17 people and similar results were observed by them for articles published in other disciplines such as chemistry. Thus, from these observations one must (mistakenly) conclude that scientific articles are not well read and that nearly all articles distributed in journal issues are not read, resulting in a huge waste of paper.
20. The problem is that these results represent only the sampled responses. That is, these sample data were not extrapolated to the entire sampled population. A careful reading of their report to NSF shows that the amount of reading of psychology articles is estimated to be 520 readings per article. Even the estimated average of 520 readings per article is an underestimate, because articles are often initially read (and reread) beyond the two-month period following publication and an appreciable additional amount of reading is from separate copies of articles. In fact, further study of the reading by psychologists suggests that the total should be closer to 860 readings per article and about 13 percent of articles distributed by psychology journal personal subscriptions are read. Furthermore, information read by a scientist is often passed on to colleagues through word-of-mouth, thus extending the usefulness and value of this published information.
21. Other studies which replicated the Garvey and Griffith survey method produced even higher estimates of readings per article. Fritz Machlup, an eminent economist, estimated that articles from eight economic journals were read an average of 1,240 times shortly after publication (Machlup & Leeson 1978). A King Research survey of cancer researchers showed that they read Journal of the National Cancer Institute articles an average of 1,800 times per article over the life of the articles.
22. Our current estimate of the amount of reading of U.S. published scientific articles is about 900 readings per article. A series of surveys of scientists from 1977 to 1998 show that the average amount of reading per scientist has not changed much over the years. A national survey done for NSF in 1977 showed that university scientists had an average of 150 readings per scientist per year and the amount observed from a university survey in 1993 was 188 readings. Scientists working elsewhere had averages of 90 readings in 1977 and 106 readings in surveys conducted from 1994 to 1998. A "reading" in all surveys is defined by us as "going beyond the table of contents, title and abstract to the body of the article". It is cautioned, however, that the number of readings is not the same as number of articles read because scientists can read an article many times. The total annual amount of time spent reading has not changed much over the years, although estimates of the reading time spent per article has increased from about 45 minutes per article to 52 minutes, perhaps because the average length of articles increased from 7.4 pages to 11.7 pages during this time.
23. Publishing delays are far longer than authors and readers would like. There was a downward trend in publishing time during the 1960s and 1970s. In 1977, a national survey showed an average publishing delay of ten months (with psychology articles being 20% higher; these articles had the highest rejection rate among eight scientific fields). However, this time then seemed to increase on average until the mid1980s, due in part to the burden of there being more articles published per journal. Some evidence suggests that delays then started to back down partially because of an increase in the use of electronic processes by publishers in the 1980s. Much of the delay is due to manuscript selection, refereeing, and editing. While there are cynics, there seems to be a general agreement among both authors and readers that these processes contribute to the usefulness and value of scholarly journals. (Some of the objections come from scientists engaged in "big science" projects who say that these processes are already adequately achieved in the grant-funding process and in their labs prior to article submission.)
24. A question arises as to the importance of speedy publication. One way to assess the importance of publishing speed to readers is to examine the reading of recently published articles in terms of their usefulness and value. About three-quarters of articles read from recently published journals are found by browsing, which tends to be done to keep up with the literature. This reading is also done more for teaching purposes than for research. Nevertheless, these readings of recently published articles are rated high in importance for these purposes and scientists spend an average of about 43 minutes per reading, which is an indicator of the value placed on the information read. Presumably, getting information earlier would indeed help. It is emphasized, however, that most reading of older articles is for research purposes; the older information is rated higher in importance and more time is spent reading as the age of articles increases (thus indicating greater usefulness and value).
25. While evidence suggests that amount of reading and time spent reading have been relatively stable over the past 20 years, there have been some changes in the ways in which scientists identify the articles they read and there are appreciable differences in the sources of these articles. Surveys from 1993 to 1998 show that scientists identify articles they read by browsing through journal issues or bound volumes (62% of readings are identified in this way), by automated searches (12%), by having other persons tell them about the articles (11%), by using citations found in other articles, books, etc. (9%), or by other means such as current awareness services, printed indexes, and so on (6%). The extent to which these means of identification are used differ from that observed in the 1977 national survey, where identification from citations and printed indexes was higher, but browsing lower and automated searches were used for less than one percent of articles read.
26. In the 1993 to 1998 period, the scientists surveyed averaged about 120 readings of scholarly articles per year. Most of these readings came from library subscriptions (55%), with the rest from personal subscriptions (27%) and separate copies of articles (18%). The proportion of readings from the three distribution means has changed dramatically since 1977 when only 15 percent of readings were from library subscriptions (compared with 55% now), 68 percent were from personal subscriptions (vs. 27% now) and 17 percent were from separate copies (vs. 18%.) Thus, reading has shifted from personal subscriptions to library-provided journals, due in large part to a decline in the number of personal subscriptions and to better library services.
27. There are a number of factors that influence information-seeking and reading patterns. Variation among scientists' communication patterns is partially attributable to personal characteristics such as one's discipline, level of education and experience, and general communication capabilities. There are situational factors as well, such as size of organization, level of research funding, amount of discretionary funds available for information services, and availability and access to library services. However, holding such variables constant, scientists have exhibited a sound economic rationale for choosing from among distribution means (and media).
28. Two dominant economic factors which influence scientists' choice among distribution means are the "price" paid for the information and the "ease" of access and use. The price paid includes both (1) the money paid for a subscription or fee paid for a separate copy and (2) scientists' time required to identify, locate, and obtain an article. Typically scientists pay 5 to 10 times more in terms of "paid" time to obtain and read scholarly articles than in actual dollars paid for the information. Ease is somewhat different in that it involves the intellectual and physical effort necessary to access the information. On the other side of the equation is what is purchased, that is, the information content and attributes such as quality, accuracy, conciseness, and so on. These value-added attributes are, of course, what publishers contribute through manuscript selection processes, arranging refereeing and editing.
29. In a given year scientists read at least one article from an average of 18 scholarly journals. However, they tend to read only a few of these journals extensively and most of them sparsely. For example, across all journals read by scientists only five percent of them are read more than 25 times by a scientist (on average) and about 80 percent are read less than 10 times. Said in another way, about nine of the 18 journals (read at least once by a scientist on average) are read less than five times; about 14 of the journals are read 10 times or less, and only one journal is read more than 25 times. The amount of reading of a journal has a major bearing on whether it should be purchased, depending, of course, on the price compared with the cost of using alternative sources of the article. In the past, libraries have been the principal alternative to purchasing journals.
30. Considering journal prices and costs to scientists in terms of their time, one can establish a break-even point in the amount of reading at which one should subscribe or go to a library to read. Based on the probability distribution of journal readings, it is less expensive to subscribe to two or three of the 18 journals mentioned above if priced at $100, but only one of them if priced at $200, and very rarely (0.8% of journals) if priced at $500. Thus, the personal subscription demand is found to be extremely sensitive to changes in price. When scholarly journal prices began to escalate in the 1970s, the number of personal subscriptions began to drop. In fact, over the past 20 years, the average number of personal subscriptions dropped from an estimated 5.8 subscriptions per scientist to 2.7 subscriptions (these being mostly subscriptions that are part of society membership). This decline in circulation also resulted in the fact, as mentioned earlier, that the proportion of reading from personal subscriptions has declined from 68 percent in 1977 to 27 percent in the 1993 to 1998 period.
31. Age of the articles read is an aspect of reading patterns, and information-seeking patterns are affected by the age. Studies over the years show that most reading is of currently published articles, but the amount of reading of articles decreases gradually over a long period of time. Two study results, observed about 35 years apart, are displayed in Table 1:
Age of Article Proportion of Readings (%) Read (Years) Operations 1993-1998 Research Group (1960)
<1 61.5% 64.4%
1-2 13.3 14.3
2-3 2.6 3.1
3-5 8.4 6.0
5-10 10.2 4.8
10-15 1.7 2.3
>15 2.3 5.0
It appears that the age of articles read has not changed much over the years.
32. Age of articles read is an important aspect of reading patterns because very few electronic journals have been established for more than five years. When electronic journals become prevalent, there is a concern that scientists will neglect to search for older articles and when they do, that they will not be able to locate and/or retrieve them because they are not available electronically. As mentioned earlier, the older articles tend to be used for research purposes and, on average, they are found to be more useful and valuable than the recently published articles. Thus, older articles should not be ignored in the future by scientists or by electronic journal system designers.
33. As one would expect, few articles over five years old are found by browsing. Citations are the most frequent means of identifying older articles (about 36% of articles read) with automated searches, other persons, and other means used about equally. Very few articles over five years old are read from copies of personal subscriptions (about 2%) or from other sources (4%), but rather scientists rely on their library to obtain the older information (about 94% of these readings). Some of these library-provided articles had been read by the scientists before (about 32%) perhaps to keep abreast of the literature or for teaching purposes, but then a later research need for the information arose.
34. A substantial subterranean flow of separate copies of articles has flourished in the past 50 years. The 1977 national survey showed that scientists received 38 million copies of articles in preprints (2 million), reprints (27 million), and photocopies given to users by interlibrary loan (4 million), colleagues (3.5 million), and authors (1.5 million). Psychologists received about two million copies of articles in these ways. For all scientific fields the total increased to an estimated 47 million in 1984 and is currently expected to be over 100 million. The proportion of readings from separate copies of articles has not changed, but reprint distribution has decreased and interlibrary loan (ILL) and document delivery together have increased tenfold from four million in 1977 to well over 40 million today. This increase is due in large part to library subscription cancellations and improved ILL and document delivery services. Once digital databases include older articles, this separate copy distribution is likely to be overtaken by online access. Certainly, the success of Ginsparg's Los Alamos National Laboratory preprint and archive system, the ACM Digital Library, and other such systems seems to confirm this eventuality.
35. Availability of separate copies of articles is an essential aspect of the scholarly journal system. It provides relatively inexpensive alternatives to journal subscriptions when a journal is infrequently read by a scientist or collectively by a library's users. This presents a stabilizing factor in the system. If the cost of a subscription (i.e., the cost per use) becomes too great, the purchaser can use the separate copy alternative. On the other hand, if the cost of obtaining separate copies of the articles of a journal becomes too great, one can purchase the journal or go to an even different source such as the author or a colleague.
36. Clearly, availability of online access to articles will affect the tradeoffs and dynamics of the system. However, there will undoubtedly continue to be instances when a paper-based subscription will cost less per reading (or be preferred for other reasons) than either an electronic subscription or access to separate copies, times when electronic subscriptions cost less (or are preferred), and instances when separate copies are preferable. One should not think of the availability of separate copies as being a replacement, but rather one alternative in a mix of alternatives that are best for scientists.
37. Characteristics of a sample of 715 U.S. scientific scholarly journals were tracked from 1960 to 1995. During this period the total number of journals increased from an estimated 2,816 in 1960 to 4,175 in 1975 and 6,771 in 1995, an increase of 140 percent over this period of time. Generally, from 1975 to 1995 journal sizes increased in number of articles (85 to 123 articles per journal) and number of article pages (630 to 1,434 pages per journal). A subsample of about 40 psychology journals shows that they tend to be smaller than journals in other fields and they were observed to decrease from 1975 to 1995 in terms of average number of articles per title, but increased about 75 percent in number of pages, therefore also indicating longer articles on average.
38. Average circulation for scientific journals decreased from 6,100 subscriptions per title in 1975 to 5,800 subscriptions in 1995, but the median decreased from 2,900 to 1,900 subscriptions, thus signifying a highly skewed distribution in which a few journals have large circulation and most have small circulation. These data also show that "the rich are getting richer and the poor poorer" in terms of circulation. Psychology journals appear also to have undergone a decrease in average circulation. Part of the decline in circulation of scientific journals is due to a dramatic 7.3-fold increase in journal prices. Psychology journals appear to cost less on the average, due in part to their smaller size, and their price increase was observed to be somewhat less than journals from other fields which saw a sixfold increase.
39. There are many reasons for spiraling prices since the 1970s, including inflation and the increased size of journals. However, these two factors account for just over one-half of the price increases and little of the increases appear to be due to an unusual rise in cost of resources used in publishing (e.g., labor, paper, etc). In the 1970s several events and phenomena seemed to have triggered some large price increases: unusually high inflation, fluctuating international exchange rates, relatively small increases in library budgets, and less discretionary funds available to scientists for purchases.
40. Since personal subscriptions are much more sensitive to price changes, they began to decrease rapidly first. To make up for lost revenue, publishers increased prices to libraries at an accelerated rate. The nature of publishing costs is that they involve high fixed costs and relatively low reproduction and distributing costs (i.e. typically about $30 to $40 per annual subscription). Thus, low circulation journals, say under 2,500 subscribers, have higher costs per subscription, and therefore prices, than high circulation journals. For example, a typical journal with 500 subscribers costs about $775 per subscription, but only $180 with 2,500 subscribers. As mentioned above, many more journals have fewer than 2,500 subscribers now than in 1975, requiring higher prices in order to recover costs.
41. The unfortunate aspect of spiraling prices is that the overall cost of the scholarly journal system has remained relatively constant on a cost- per-reading basis. The cost of the entire journal system is $45 billion or about $60 to $70 per reading (i.e., with 87% due to authorship and reading, publishing about 7%, and libraries and other intermediaries about 6%). In current dollars, the 1977 cost is about $16 billion or about $50 to $60 per reading. The small increase in cost per reading from 1977 to recent years is due to an increase in scientists' time used to obtain and read articles. Both publishing and library costs per reading have decreased slightly.
42. Since their birth in the 17th century, scientific scholarly journals have become the most used type of publication and, for most fields of science, the single most important channel of communication. Over the past 40 years, numerous studies indicate that journals are extensively read; the information they contain is extremely useful for research, teaching and lifelong learning; and the information is valuable in terms of the favorable outcomes from its use. All of these indicators have remained stable over the years. Furthermore, taking into account the number of journal system participants (i.e., authors, readers, publishers, and libraries and other intermediaries), the overall system costs have remained relatively constant, though they have increased slightly due to the increased costs of scientists' time necessary to obtain and read scholarly articles.
43. However, despite overall stability over the past 20 years, there has been considerable turmoil within the scholarly journal system:
Journal prices have increased far more than one would expect based on inflation and their increase in size.
Personal subscriptions have decreased from an average of 5.8 subscriptions per scientist to 2.7 subscriptions.
Thanks to the growth of the number of scientists and reduced personal subscriptions there are 19 million fewer personal subscriptions than might be expected, resulting in billions in lost revenue to publishers.
Library subscriptions are shown to be relatively insensitive to price increases. This is due in part to the comparative cost of the principal alternative (interlibrary borrowing/document delivery) remaining prohibitive for many well-read journals (i.e., about $20 to $30 per article).
Libraries have reduced book and other acquisitions and cancelled some journal subscriptions to cover the increased prices of journals.
Library journals that are priced high and infrequently read tend to be cancelled, and separate copies of needed articles obtained by interlibrary borrowing or document delivery.
The number of articles obtained by interlibrary borrowing/document delivery has increased from about four million in 1977 to over 40 million today.
Separate copies from preprints, reprints, and copies provided by colleagues and authors have increased from 34 million to about 60 million.
Scientists read much more from library-provided journals (i.e., 15% of readings in 1977 to 55% now).
Scientists, on average, spend more time obtaining articles, due in part to using their libraries. They also spend more time reading articles.
Individual scientists read from about 40% more scholarly journals now than 20 years ago. This is thought to be because they have been able to broaden searching through automated searches and because there is greater access to separate copies of articles.
44. These trends have meant that scientists spend more of their scarce time obtaining and reading articles; libraries spend more, but provide less information to their users; and publishers are blamed for these trends and lose considerable potential revenue. Thus, all participants appear to be losing, even in light of the overall stability in system costs and the use, usefulness, and value of scholarly journals.
45. There has been tremendous interest in and growth of electronic journals and digital full-text article databases. Many believe that new technologies mean that flaws in traditional paper journals will be eliminated and, perhaps, scholarly journals will become irrelevant. However, the persistent use, usefulness, and value of scholarly journals suggest that one must tamper with this communication channel cautiously, and not risk its value to scientists merely for the sake of change.
46. As scholarly journals move more into electronic publishing and the establishment of digitized text databases, pricing is going to be perhaps the most important issue that must be faced. Electronic journals will not alleviate the pricing problem because most publishing activities and corresponding costs are common to both traditional paper-based and electronic media. The small savings from eliminating printing and mailing are offset by increased investment in and costs associated with technology. Furthermore, paper-based subscription costs (per reading) become competitive with electronic subscriptions if the journal subscription is read enough.
47. There is a tendency to think in black and white terms; that is, to think that scholarly journals must be in paper-based or in electronic medium. Yet the complex array of multiple communication channels, distribution means within channels, and varied media used for each distribution means suggests that each combination has found its niche based on costs, user needs and user requirements. The infusion of new technologies, electronic publishing, and digital databases will profoundly perturb the overall communication system in general and the scholarly journal system in particular. However, evidence suggests, at least in the short term, that electronic journals and digital full-text article databases will clearly find a niche, but not necessarily replace the traditional paper-based medium. New pricing policies will undoubtedly determine what the niche will be. Site licensing, differential pricing, and the potential of new sources of revenue (particularly to cover fixed article processing costs) need to be explored in order for electronic publishing and digital databases to meet their full potential.
48. The intrinsic value of electronic journals and complementary digital databases is that they introduce additional options from which scientists and libraries can choose. Based on supply costs/prices:
New journals and journals serving small disciplines and audiences will tend to be exclusively electronic journals.
Established journals with large circulations will tend to be made available in both paper-based and electronic media. Subscribers (i.e., scientists, libraries) will choose one medium or the other depending on the extent of reading and on personal preferences.
Access to separate copies of articles from digital full-text article databases will replace and expand on the current paper-based distribution of separate copies of articles. However, it is unlikely that this phenomenon will replace journal subscriptions for some time to come, if at all.
Libraries will continue to serve scientists with less emphasis on parent organization storage (and archives) and greater emphasis on identification, location, and physical access to needed information wherever found.
Based on demand/costs, scientists and libraries will continue to make economically rational choices among alternative distribution means and media, just as they have in the past.
49. Scientists' time and effort are important considerations in the future development and implementation of technology enhancements to electronic journals and digital text databases. There is overwhelming evidence that scientists take into account their time and other resources in their information-seeking: use of libraries is highly correlated with distance to the library; scientists have replaced the high cost of cancelled subscriptions by spending more of their time on reading library copies and studies have shown that distance to the library is related to the number of personal subscriptions (i.e., those further away subscribe to more journals); scientists in industry say that they use information specialists to conduct difficult automated searches because it saves their time (and specialists can do the search better and faster); and summed across all such services scientists report a substantial amount of time saved each year.
50. There has been considerable "hype" concerning electronic journals and digital databases, some prognostication that is unlikely to be achieved. However, new technologies have opened the way for considerable improvement that can be accomplished in the future. For example, to name but a few innovative electronic services: multimedia; hyperlinks within and among documents; reducing redundancy among channels; providing options for different levels and types of information (e.g. multiple journals, single journals, articles, titles and abstracts, sections paragraphs, citations, backup data); providing a trail of modifications and updates; improving organization, control, identification, and retrieval; introducing multiple article reviewing and rating methods; establishing article usage data; disseminating groups of articles selectively. However, these improvements may take considerable time to achieve acceptance, if ever.
51. Authors will undoubtedly continue to choose the scholarly journals that meet their needs, but they should also consider whether or not readership of these articles might be affected by journal prices, recognizing that prices must be high for journals serving small specialties. Authors who choose to publish largely on their own Web sites need to recognize that readership, and therefore value, will be affected if their articles are not input to a reputable bibliographic database. Authors should also be aware that their greatest readership may be outside the academic community, especially if their articles are in the physical and life sciences. Peer review and editing are still important to readers and will undoubtedly continue to be sought by authors and readers. One important aspect of authorship is that the discipline necessary to put one's thoughts on paper, particularly with the knowledge that the manuscript will be critically reviewed, often actually enhances creativity.
52. Research over the years continues to demonstrate that scientific journals have tremendous use, usefulness, and value. This will undoubtedly continue to be the case regardless of whether articles are published in paper-based or electronic media, as authors and readers recognize that electronic publications provide an alternative, not necessarily a substitute, for traditional paper-based journals. Readers need not be put off by screen displays because in-depth reading can be read from printouts which are relatively inexpensive. Some of the most useful and valuable scientific information is found in older articles that typically are not yet available electronically and therefore require an alternative, more traditional means of identification and acquisition.
53. For research and teaching purposes, it is important not to rely solely on Internet sources of information, which do not provide some quality mechanisms, for either primary or secondary publications. Online searches are often best conducted on well-established databases because of their completeness and quality, preferably those sanctioned by a professional society (e.g., American Psychological Association) or that are well established such as Science Citation Index. Librarians or information specialists will continue to be a valuable resource for identifying, locating, and assessing needed information. Their service is likely to become even more relevant as information on the Internet becomes more prolific and chaotic. Users need continually to inform specialists about their information needs and requirements by utilizing specialists' strengths, including their intimate knowledge of both online and paper-based collections and how to access them.
American Psychological Association (1963/1968) Reports of the American Psychological Association's Project on Scientific Information Exchange in Psychology. Vol. 1 Overview Report and Reports 19, 1963; Vol. 2 Reports 1015, 1965; Vol. 3 Reports 1619, 1968. Washington, D.C.: American Psychological Association.
Garvey, William D. (1979) (Ed.) Communication: The Essence of Science. Oxford: Pergamon Press.
Garvey, William D., & Belver C. Griffith (1963) The American Psychological Association's Project on Scientific Information Exchange in Psychology. Report 9. Washington, D.C.: American Psychological Association.
Kessler, M. M. (1967) Some very general design considerations. In TIP system report, Appendix H. Cambridge: Massachusetts Institute of Technology.
Machlup, Fritz, and K. W. Leeson. (1978) Information Through the Printed Word: The Dissemination of Scholarly, Scientific, and Intellectual Knowledge. Vol. 2, Journals. New York: Praeger.
Operations Research Group (1960) A Case Institute of Technology. An Operations Research Study of the Dissemination and Use of Recorded Scientific Information. Cleveland, Ohio: Case Institute of Technology.
Tenopir, Carol, and Donald W. King (2000) Towards Electronic Journals: Realities for Scientists, Librarians, and Publishers. Washington, D.C.: Special Libraries Association. http://www.sla.org