It has been recognised for many years (Ziman, 1968 ; Ravetz, 1973; Meadows, 1980 ) that the dissemination of information is not the sole function of the scholarly journal literature. The paradoxical survival of this apparently archaic form of literature has depended on its multiple functions, which are essential to the orderly functioning of a scholarly community. The four chief functions are:
The first of these has not necessarily ever been the most important. Occasionally a very important new finding has been reported by means of a rapid communication such as a letter to Nature ; for example, the famous paper by Watson and Crick (1953) was published 23 days after submission. But this is not the usual route for less epoch-making discoveries. Usually these will be reported initially at a conference; even earlier, they may have been discussed informally among workers in the field. The informal channels have been referred to as "the invisible college" (Crane, 1972 ). Historically, the invisible college worked through international telephone calls, travel to other universities, and conference attendance, and therefore participation generally required money. As a result, membership of the invisible college tended to be confined to senior scholars; junior members of their research teams participated vicariously through their boss, and people not situated in major research departments tended to be excluded. The Internet has transformed this situation; the invisible college has been democratised and reinvigorated, and academic life radically changed, by the almost cost-free discussion across time and space that electronic mail offers.
It is important , however, to distinguish between academic debate and scholarly publishing. Opinion is free but facts are sacred. It is important that we maintain an unchangeable archive of verified research results. If I drive my car across a high bridge, I need to be confident that the engineers who designed the bridge had access to sources of thoroughly reliable and tested information about the properties of materials, the geology of the foundations, and so on. If I go to my doctor with a life-threatening illness, I need the doctor to have reliable medical-research information available, so that the best possible treatment can be prescribed. A nice example of this important point has recently come to my notice (Sokal, 1996). Alan Sokal, a physicist and with unimpeachable left-wing political credentials, submitted a paper with an impeccable reference list to a left-wing journal of postmodernist social science, Social Text. The paper was in fact a parody, but the editors did not notice this because the views it expressed coincided with their own. Sokal then admitted the deception in a paper in another journal, Lingua Franca. His purpose was to emphasise the importance, in the sciences at least, of scientifically demonstrated facts, which are not replaceable with culturally defined beliefs, even the one that one holds oneself.
Hence the second and third functions of the scholarly journal (quality control and the preservation of the archive) can be seen to be essential not only to good scholarship, but also to safe practice in the professions. There has also been debate about the potential of the net for collaboration, and Rzepa (1995) has coined the word "collaboratory" for the virtual space in which scientists at remote sites can work together. There has, in my opinion, been a lack of clarity in much of the electronic publishing debate about the difference between discussion (which the Internet facilitates wonderfully) and the recording of research results.
It is also important that workers in any field of human activity should be encouraged to do thorough, rigorous work, and not cut corners. For research workers in particular, both careful attention to appropriate methodology and an impartial and sceptical view of their own results are desirable. Writing up their results and making them available to others (including non- researchers such as practitioners and students) is also an important part of their professional duties.
It has recently been reported (Major, 1996 ) that, in order to secure patent rights in the USA, scientists in the UK will have to write up their daily work according to strict rules in bound, not looseleaf, notebooks, and certainly not electronically. Similar levels of certainty are desirable even in work that is not destined to be patented.
The scholarly journal literature has always provided the mechanism by which researchers are given an incentive to be careful and rigorous in their work. Peer review is not a simple binary decision, accept or reject. Scholars were aware of the "pecking orders" of journals in their fields long before Eugene Garfield (1977) started to document them through his citation analysis studies. They submit their better work to better journals, and work they know to be less distinguished to those lower on the pecking order. Thus there has evolved a graduated scale of quality facilitated by the journals. Reputation and status depend on publishing in good journals, and further research funding and job promotion opportunities depend on this reputation. The necessary incentive to quality is provided through the journals system. This is not corruption, but a functioning and effective (though not necessarily efficient) method of allocating research posts and resources.
Thus it is quite understandable that the fourth function mentioned above (recognition of authors) should be seen by researchers as the most important function of academic journals. There is no real evidence that the number of papers per researcher has increased dramatically: the number of papers has increased because the number of researchers has (Ziman, 1980; Meadows, 1993). Anecdotal evidence, however, suggests that there has been a tendency towards "salami publishing" - slicing up one's research results thinner and thinner in order to get more publications from one research grant. Some people believe that "more means worse", in this as in other fields, but that is a matter of opinion. The problem is that universities have grown, and institutions formerly not active in research have started to compete for research funding, while the budgets of university libraries have in many cases fallen in real terms. Thus universities are producing more research results needing to be published, but their libraries lack the resources to buy the resulting publications.
There always has been some scientific fraud (Lennon, 1996 ), and the journals' peer-review system has not always uncovered it. There are dishonest people in every profession, but I would contend that no specific corruption in the journals exists. The era of the widespread launch of new journals is now over; the reduced purchasing power of libraries has ended it. Well-established journals will seek to retain or improve their place in the pecking order, not jeopardise it. In the competition between printed journals, and between them and newly founded electronic journals, a reputation for quality is the main asset that a journal possesses. It has a strong incentive to maintain that reputation.
The library funding crisis could in principle find a solution through the technology of the Internet. We need to ensure, though, that all four functions of the academic journal are provided for, especially the quality-control and canonical-archive functions. McKnight (1995)  has pointed out that as a very minimum the electronic journal must provide for all the functions that the printed journal performs; preferably, of course, it should provide for new features, such as multimedia content and rich hypertext linking, as well. But the preservation of good scholarship is more important than new bells and whistles, any one academic's career, or any one publisher's profits. Everyone will suffer if there is not a corpus of information that can be relied upon.
Harnad (1996), Odlyzko (1996) and others have suggested a new mode for "esoteric" publishing that avoids commercial motivations completely, and uses the Internet for direct transmission of information from scholar to scholar without middlemen. Insofar as some research communities really are small, they probably are right. But in many fields there is a large potential audience, and there are readers of scholarly papers who are not researchers - medical practitioners are an obvious example. Harnad (1996) has conceded that commercial publishers will probably continue to be involved with those journals that have larger, non-research markets. Central to my argument, however, is the contention that electronic-only journals, whether paid for or free, must have quality-controlled content that, once published, cannot thereafter be altered by the author or anyone else, other than by a corrigendum that also passes through the editor. This in no way underestimates the value of the Internet for scholarly debate, but it recognises that there is more than one kind of communication that is necessary if good scholarship and good practice in the professions are to be encouraged.
It is true in theory that all the top researchers in a field could stop submitting their articles to commercial journals and refuse to referee for them, and transfer their energies to new electronic journals, thus raising their prestige. In practice it is unlikely that this will happen by voluntary action. This has led some academics to suggest that universities should retain the copyright in work done by their staff and publish it themselves instead of submitting it to journals published by outside (especially for-profit) publishers. This is not wrong in principle so long as the work is peer-reviewed. I have argued elsewhere (Rowland, 1996 ) the case that large journals will always need professional staff for administrative and subeditorial duties and that hence they cannot be free, unless subsidised. Furthermore, the owners of the titles of major journals, which carry prestige, are not going to relinquish them readily.
Most of the major scholarly publishers are now offering their journals in parallel print and electronic forms. I believe that the commercial publishers will come under intense pressure to use the new technology to reduce their prices. The current position where the electronic version costs the same as or more than the printed is untenable, given the financial position of the libraries that are their major customers. The large publishers, both for-profit and not-for-profit, argue that ceasing to print their journals will save only a small proportion of their total budget, and that in the intermediate period when publication is in dual form, their costs are actually raised by the need to provide both forms. The radicals who argue for a complete reform of the system use zero-based budgeting to demonstrate that costs could be drastically lowered, if new electronic-only journals were started from scratch.
It is hard to predict where this argument will end. However, the library budget crisis combined with the increased determination of universities to retain the intellectual property rights in their research seem likely to require some compromise from publishers. The HEFCE Pilot Site Licence Initiative (PLSI) (Bekhradnia, 1995 ) offers one possible way forward, offering as it does electronic access free at the point of use to all members of a subscribing university. It also demonstrates that the UK government is likely to support the survival of a profitable private-enterprise export industry. But the smaller journals, genuinely "esoteric" seem likely to be replaced by free Internet publications on the Harnad model, to the detriment of publishers who have specialised in low-circulation, high-priced scholarly journals.
Reply received Monday, 27th January:
I find Mr Rowland's article both timely and relevant, as I have lately been assuring my colleagues 'In 10 years there will be no scientific journals, only Web sites.'
The matter of archive integrity is interesting. If Mr Rowland's bridge engineer got his information from a paper containing errors, perhaps known to the author but awaiting publication, would he not prefer a web site which can be updated and corrected in response to new data?
The practical question for libraries is: how much larger will our buildings need to become, say in 50 years, if we do not move to a new system of publication?
Library Systems Manager
John Rylands University Library, Manchester firstname.lastname@example.org
 Ziman, J.M. (1968) Public Knowledge: The Social Dimension of Science, Cambridge University Press, Cambridge, UK.
 Ravetz, J.R. (1973) Scientific Knowledge and its Social Problems, Penguin Books, Harmondsworth, Middlesex.
 Meadows, A.J. (Ed.) (1980) Development of Science Publishing in Europe, Elsevier, Amsterdam.
 Watson, J. D. and Crick, F. H.C. (1953) Molecular structure of nucleic acids: A structure for deoxyribose nucleic acid. Nature, 171 (4356), 737-738.
 Crane, D. (1972) Invisible Colleges: Diffusion of Knowledge in Scientific Communities, Chicago University Press, Chicago, IL.
 Sokal, A.D. (1996) Transgressing the Boundaries: Towards a Transformative Hermeneutics of Quantum Gravity, available at: http://www.nyu.edu/gsas/dept/p hysics/faculty/sokal/index.html
 Rzepa, H.S. (1995) The future of electronic journals in chemistry. Trends in Analytical Chemistry, 14, 464.
 Major, L.E. (1966) Making it patently obvious. The Guardian, Higher Education section, 7 October 1996, vi.
 Garfield, E. (1977) What is the 'core' literature of biochemistry as compared to the 'core' of chemistry? (Journal citation studies, 1), in Essays of an Information Scientist (2 vols), ISI Press, Philadelphia, PA, vol. 1, pp. 262-265 (and many subsequent Journal citation studies in the same book). >
 Ziman, J.M. (1980) The proliferation of scientific literature: A natural process. Science, 208, 369-371.
 Meadows, J. (1993) Too much of a good thing? Quality versus quantity, in The International Serials Industry (Woodward, H. and Pilling, S., Eds), Gower, Aldershot, pp. 23-43.
 Lennon, P. (1996) Cooking it up in the lab. The Guardian, Online section, 10 October 1996, 16-17.
 McKnight, C. (1995) The human factors of electronic journals, in Project ELVYN; An Experiment in Electronic Document Delivery (Rowland, F., McKnight, C. and Meadows, J., Eds), Bowker-Saur, London, pp.37-47.
 Harnad, S. (1996) Implementing peer review on the net: Scientific quality control in scholarly electronic journals, in Scholarly Publishing: The Electronic Frontier (Peek, R.P. and Newby, G.B., Eds), MIT Press, Cambridge, MA, pp. 103-118.
 Odlyzko, A.M. (1996) Tragic loss or good riddance? The impending demise of traditional scholarly journals, in Scholarly Publishing: The Electronic Frontier (Peek, R.P. and Newby, G.B., Eds), MIT Press, Cambridge, MA, pp. 91-101.
 Rowland, F. (1996) The need for management of electronic journals, in Scholarly Publishing: The Electronic Frontier (Peek, R.P. and Newby, G.B., Eds), MIT Press, Cambridge, MA, pp. 243-250.
 Bekhradnia, B. (1995) Pilot national site licence initiative for academic journals. Serials, 8( 4), 247-250.
Fytton Rowland is a lecturer in the Department of Information and Library Studies at the University of Loughborough