Published in Learned
Publishing, Vol. 12, No. 2, April 1999. (ISSN 0953-1513)
Copyright - ALPSP 1999
The limitations of existing proposed and operational publishing models intended to replace the academic journal are briefly reviewed. Three 'insights' are described, the first is into the 'means/end' confusion of much current net-based publishing activity, the second is that it is the purpose, not the form, that is the important aspect of the traditional STM journal model, and the third is that satisfactory net-based publishing models need not contain a central publisher. From this new viewpoint is developed the Deconstructed Journal model which it is suggested is a better model for network based academic publishing. It also solves some of the problems of the current model. Although the main focus of the DJ model is the replacement of the STM journal it has implications for all areas of academic journal publishing.
The proposed publishing model described in the latter half of this article has been developing in my mind for over five years. The beginnings of this model can be traced back to an idea for network-based scholarly publishing which I described at a meeting held at the Royal Society in 1993 . I also described variations of this early model to other meetings in the UK  and India .This article itself is an extended and updated version of a short paper given at the ICCC/IFIP Conference on Electronic Publishing in 1997 (Smith, 1997).
Initially I did not feel it was worth writing about that early model as I felt it was so obviously a candidate to replace the traditional paper-based model that others must already be describing it in the literature of library and information science. However, this has not been the case. Although other models have been proposed, and tried, there has not to my knowledge been any model described quite like the one I currently call the 'Deconstructed Journal' (DJ) model. The nearest I have seen is in a paper by Ginsparg (1996) where he writes:
"Any type of information could be overlayed on this raw archive and maintained by third parties." ... "the average reader could benefit from an interface that recommended a set of "essential reads"...."
However he is concerned with the problem of accessing a single large archive source and does not extend the idea he is describing to distributed sources. Neither does he point out that it has a close resemblance to the 'recommending' or 'filter' role of the traditional journal. Although the ideas expressed here could apply, to a greater or lesser extent, to most areas of academic journal publishing I have chosen to make the STM journal the main focus of this article. The STM journal has a more clearly defined role for its authors and readers which makes it easier for me to compare the old and new models discussed. So from this point on I will use the phrase 'STM journal' rather than the more general 'academic journal'.
Another very recent development that has echoes of the DJ model is the New Journal of Physics, IWR (1998), launched by the Institute of Physics (UK) and the Deutsche Physikalische Gesellschaft (Germany) in the autumn of 1998 . The relationship to the DJ model in this case is the proposal that the author pays for 'publication' and the end product is free to the reader.
There have been two distinct approaches to moving scholarly publishing to the net.
The first approach leads to models that take advantage of the capabilities of the net but usually do not have the formal structure required for the solid, trustworthy transfer of organised knowledge. Sometimes they have been adapted to support such features as peer review or other forms of content evaluation but this has often been 'bolted on' afterwards rather than being intrinsic to the design.
The second approach limits the capabilities of the net as a knowledge transfer medium by forcing on it concepts that make no sense in the network world. For example we still have the continuing existence of the 'page' and the 'issue' in e-versions of STM journals. At a deeper level the second approach also transfers to the net many of the bad points of the paper based model. These include: cost of subscriptions, loss of IPR by the author, restricted access to publication due to lack of space, etc. Finally the centrist approach of the current model does not fit the basically distributed model of the net.
For examples of the variety of models proposed for academic e-publishing see Bailey (1994), Ginsparg (1996), Harnad (1990, 1996), McKnight (1993), Odlyzko (1995), Piternick (1991), Rowland (1994), Savenije (1997) and Swinnerton-Dyer (1992).
Swinnerton-Dyer (1992) proposed a publishing model heavily influenced by the paper model in operation and based around a centralised server. Bailey (1994) briefly reviews selected proposed large-scale models and then lists a set of basic requirements he feels any e-publishing model must fulfil. Odlyzko (1995) includes an outline model based on the interactive and less formal possibilities of net-based publishing. Harnad (1990) is concerned mainly with the pre-formal communication stage between researchers, but Harnad (1996) elaborates the earlier model to include a mechanism for peer-review. Ginsparg (1996) describes the operation of a real working e-publishing model. In this case the question is not whether it works (use of the Physics E-prints Archive is currently running at around 70,000 connections per day  ) but whether the implicit model it embodies is applicable to other areas of research. Savenije (1997) argues an opening case similar in some ways to mine but his analysis of the publisher and/or journal roles is too simplified and his suggested model is focussed mainly on possible roles for the library.
Although the ACM Electronic Publishing Plan, (ACM, 1995), is a list of assumptions, goals and proposed actions needed to move towards e-publishing, rather than an explicit description of an e-publishing model, it does contain an implicit model. In this implicit model the concept of the journal is weakened, the issue disappears, and the idea of access to a 'database of documents' (similar to that described by Ginsparg (1996)) becomes the "core of the membership package".
For other surveys of attempts to replace the STM journal see Bailey (1994), Lines (1992) and Piternick (1989).
Given this plethora of competing (and in some cases operational) e-publishing models why is there a need for yet another model?
The problem with most of the current network-based attempts to replace the paper STM journal (and many of the proposals for future models) is that they have set out to mimic or replicate the existing mechanisms without asking the fundamental question 'Why are they the way they are?'. In essence they have confused the current 'means', the STM journal, with the real 'end', scholarly intercommunication and recognition.
However, before dismissing the paper-based journal and the model it embodies there is something it can teach us. The current model of the STM journal is the result of many years of evolution. The fact that it has retained its present general structure for nearly two centuries implies it must be doing something right, i.e., it must be playing a real role. Any replacement must therefore play the same roles or, to rephrase it more strongly, it must satisfy the same needs.
To apply this insight the first move therefore is to decide just what roles the STM journal plays.
These are the roles most users (and librarians , etc.) would say are played by the STM journal.
The first two items on this list can also be considered as a form of filter which allows through only information that is relevant and of sufficient quality. This idea will be considered more fully below in the context of the proposed model.
In addition to the obvious roles played by the journal it also has hidden or non-obvious roles which are nevertheless important to the academic research community.
We will return to this analysis of main and hidden roles when we consider the proposed new model.
Reviewing the above 'analysis' of the roles of the STM journal without any commitment to the current model (in fact with a positive desire to escape the limitations of the current model) the one thing that is obvious is that most of the activities involved are independent (e.g. quality control activities are not concerned with distribution) and therefore there is no obvious need for these roles to be carried out, or co-ordinated, by a central organisation (i.e. the publisher). Like the 'page' and the 'issue' the need for a publisher (or central co-ordinator) is a result of the requirements of the paper-based model. As long as there was a physical item to be produced and distributed it made sense to do this as collections of papers (issues) rather than single papers. Someone (or some organisation) was needed to collect these together, arrange selection, printing, binding, and distribution. It also made sense for this person/organisation to organise the quality control (editorial board, referees, etc.) as well. Once it is realised there is no necessary requirement for a central publisher one can consider new structures (or organising models) that play the roles listed above, i.e., satisfy the needs currently satisfied by the present publishing model.
This is the essence of the third insight - the realisation that it is possible to have a model for STM publishing that can satisfy the needs of the STM community without a central publisher/co-ordinator. This can be achieved by involving a collection of co-operating actors or agencies.
Further, this distributed model far better utilises, or matches, the capabilities of the net than does any centralised model.
This basic idea of a decentralised academic publishing model designed to play all the roles of the current model based on co-operating actors or agencies can give rise to a collection of operational models which fit within its fundamental premise. What is described below, the Deconstructed Journal (DJ), is one such operational model.
I find I have needed to describe my proposed new model (the DJ) from many angles. This is because I am using words (e.g. journal, publishing, article, etc.) which may have one meaning in the old model and another in the new. This problem of words/concepts changing their meaning leads me to suspect that my new DJ model is a true 'paradigm shift' as described by Thomas Kuhn in The Structure of Scientific Revolutions, (Kuhn, 1970). Many authors writing in the area of new publishing models have described their work as incorporating a 'paradigm shift' but in many cases their new models are not new paradigms - they are simple (or complex) extrapolations or re-workings of the current model. This changing in the meaning of words/concepts during a paradigm shift is not complete proof of such an event - but it is highly indicative. To be precise it is the underlying set of assumptions derived from the three insights (most importantly the proposal that a distributed model without a central publisher could satisfactorily replace the current model) that are the new paradigm.
What follows is one way of reassembling the elements 'pulled apart' in the previous section, or viewed another way it is a way of fulfilling the roles of the STM journal using a new assemblage of actors and activities.
As one might expect, at the core of this new operational model is a Web  service similar in structure to a current subject gateway . This service contains links to relevant items of interest to its readers (who may also be subscribers). This is the Subject Focal Point (SFP) and it is the visible replacement for a specific STM journal in the current model. Some of the important differences between the SFP and current paper-based and e-journals are:
A basic version of the SFP would be created (or come into existence) in the following way:
It is possible to confuse the Subject Focal Point (which replaces, at a surface level, the journal in the current model) and the Deconstructed Journal which is the overall model within which the SFP works. This confusion can arise because in the current model most of the needs described in the analysis above are satisfied during the production of the traditional journal. In the DJ model the SFP only satisfies some of these needs and activities carried out by other organisations satisfy the rest. Therefore we cannot compare the SFP directly with the traditional publishing model we must consider all the activities described within the DJ publishing model. The following section may help explain this.
To see in detail how this model would work we need to list the various roles that need to be played by this replacement for the current STM journal model (as defined by the analysis of the roles of the traditional journal above) and indicate how, and/or by whom, the role is to be carried out within the proposed new model.
This analogy was mentioned briefly above. The realisation of the filter role of the traditional STM journal was one of the early steps that led me to formulate the DJ model. This role is the major role of the SFP where it is explicit rather than implicit as it is in the traditional journal model.
The filter role could easily be extended so that, in addition to providing links to items submitted, the SFP could have a team (like an extension to the editorial board) that actively scans the network for interesting and relevant items. This 'one-stop-shop' pointer service for all information on the net of value in a specific subject might be the major selling point for the SFP and the primary reason why a subscription is taken out. An extension of this idea might be that instead of an author submitting an item to an SFP he or she might post a reference to it in some common area concerned with the appropriate subject or subjects, knowing that 'talent spotters' from the SFPs look there for new material.
Another way of illustrating the DJ model is to consider how it might operate from the point of view of an author/researcher.
There are many possible variations. For example the article may be on a university server during the evaluation phase but be moved to a commercial or not-for-profit independent server before being 'offered' to an SFP or SFPs.
A possible reader use scenario may be:
Assuming a functionality at least equivalent to the current Web client/server environment the user will then be able to view items on screen and/or request copies. It is worth stressing that access to the primary material will be free in almost all cases. An example of where access might not be free is where an SFP recommends (i.e., points to) a commercial service if the editors considered it to be of value to their readers.
As one would expect there are many implications and possibilities that follow from this new model. A selection of the most important ones are considered here.
The scatter problem is when information pertinent to a specific area of research is spread across a number of journals. This is particularly common with emerging research areas as there may not be a specific journal or journals dedicated to the subject. It can also occur with work that straddles current research area boundaries. The DJ model completely eliminates the scatter problem. To be exact the scatter problem need not exist in the model as no published item is owned by any particular SFP - so any SFP can point to any item that the editorial board feels would be of interest to the subscribers.
A well known problem of refereeing in the current model is that authors with new ideas (especially those involving new theories or paradigms) find it difficult to get published, Schauder (1994). The DJ model, because it allows for many independent evaluators, and the possibility of grading rather than the simple pass/fail (publish/don't publish) approach of the current model, could allow unknown authors with a radical new idea to get published (more correctly 'acknowledged' or 'made visible') more easily.
Because there may be a range of evaluator organisations within a topic area it is possible the speed of throughput might increase.
Good articles sometimes miss being published because of limited space in paper-based journals. This obviously cannot happen with the SFP as it only points to items elsewhere. Obviously this is true of many other forms of e-publishing once one has escaped the physically imposed limits of the paper-based model.
A detailed study of the STM information system in the UK was carried out for the Royal Society, British Library and ALPSP in 1993, Royal Society (1993). Although this is a little dated now many of the conclusions and recommendations in the report are still relevant. The way in which the DJ model would address some of these is considered below. The numbers and heading titles referred to are taken from the Royal Society report.
Any new SFP that comes into existence can start by pointing to existing items that are relevant to the subject area in question. This again follows from the fact that no item is owned by any specific SFP.
Under the definitions of the current paper-based model the SFP is more like a 'secondary' publication than a 'primary' publication, since it points to, but does not contain, the actual information item. However it is the first point of contact, and link, between the subscriber and the information sought. In this it resembles both the primary journal and a secondary bibliographic service.
There are three main problems areas preventing easy adoption of this model. These are: community acceptance, funding, and technical. Of these the first two are by far the most important.
The major obstacle to the adoption of the proposed new model is acceptance by the user community - as an earlier writer on this topic pointed out
"the introduction of a single innovation in technology cannot be successful until it is accompanied by an appropriate set of social, behavioural, organizational, and institutional innovations." Goldhar et al (1977).
Twenty-one years later a strong proponent of e-publishing is still lamenting
"But the biggest brake on progress is still surely the reluctance of authors to entrust their work to a new, unproven, medium in place of the one that has served them faithfully for years." Harnad (1998)
Such acceptance will only come about when the professional and funding bodies accept publishing in this model as equal to paper-based or network based publishing using the traditional model. Research funding bodies in the UK are recognising network published items as long as they have passed the usual quality control tests (i.e. selection and refereeing), so one assumes they would accept the DJ model as long as the quality control was as good.
On the positive side some research areas have already diverged markedly from the traditional path without completely abandoning it. A prime example is physics and related sciences (mathematics, astronomy, etc) where the researchers are already placing their papers in the e-print archives for their colleague to access long before they are available in the traditional journals. For these groups the DJ model will not look so alien and its attributes may be more easily recognised. Indeed the New Journal of Physics (mentioned in my introduction ) includes some of the basic ideas of the DJ model. Although it is not clear from the available information whether authors will need to pass Copyright to the NJP as part of the publishing process there seems to be no reason why others could not point directly to any article within it as it is available free to any reader. To this extent it appears as a simple server/archive within the specification of the DJ model.
The second most important obstacle is finding a operational economic model that would fund activities within the new model. Although a detailed consideration is not possible here the basic funding model assumes the following:
Related to the idea of cost is 'efficiency'. By this is meant the overall cost of the system to the end-user community including the researchers and their employer organisations. At first sight the DJ model with its added actors and activities would appear much less efficient than the current model. However this is not necessarily the case. Many of the actors are in fact the same people, for example the referees are likely to be the same, so to are the academic members of the SFP editorial boards. In addition some current activities are not required, e.g., storage of the paper journals by libraries. When we add in the general efficiencies of e-publishing compared to paper publishing the differences between the two models shrinks again, e.g., the DJ has no printing or distribution costs. There are writers in the area of new models for STM publishing, for example Walker (1998), who argue that if we take into account the whole cost of the STM publishing enterprise from author to reader including the profits made by publishers and the cost of libraries (to store and make available the paper journal), the cost of e-publishing will be significantly less overall. The experience reported by Walker on the move of the Florida Entomologist (An International Journal for the Americas) to electronic form confirms this. Odlyzko (1998) also argues, with examples, that considerable savings can be made in the move from paper to e-publishing. These arguments, of course, are relevant to e-publishing in general not just the DJ model.
There are various technical problems. The following are just two of the more important ones.
Despite these problems, and others that might be encountered in the implementation of a publishing system based on the DJ model, it is felt there is no serious technical problem that cannot be overcome by the application of existing techniques.
It is proposed that the DJ model could replace the current paper influenced electronic publishing model. There are six main reasons why I feel this model is superior to most existing operational and proposed models.
Because it is based on an analysis of the roles of the time-proven traditional model the DJ model is known to be fit for it's purpose.
The DJ model far better fits the flexibility of the network as a publishing and information dissemination medium. It is a de-centralised model able to change and adapt with changes and developments in the network and it's facilities.
The DJ model solves many of the problems manifested by the current paper-related model.
The open distributed structure of the DJ model means there are openings for many new actors to play small, medium and large roles within the new model.
Because the central element of the DJ model, the SFP, does not own the Copyright of the items it points to, control of further distribution or use is left with the authors and/or their employers. This allows greater freedom for the individuals and institutions concerned to use and re-use their intellectual property. There has been an increase in resentment over they way traditional journals insist on being given Copyright as part of the publishing deal, Bachrach et al (1998), and Sutherland (1999).
The proposed DJ model has fundamental roles for such entities as the learned societies and the universities. Adoption of the DJ model or any other model conforming to the underlying assumptions on which it is based will give them more control over the STM publishing industry. This would return control to those who both produce and use the fruits of this industry.
The DJ model cannot simply evolve from the current model as there are internal conflicts, e.g., the concept of journal/article ownership in the current model versus the requirement for SFP/article non-ownership in the DJ model. Nevertheless I feel there could be a transition period during which some subject areas would set up SFPs, evaluator organisations, archives, etc, while others continued with the traditional model. The key move is the provision of trusted evaluator organisations. They are needed to provide the quality control currently lacking in most network-based publishing. Further, they can perform this role even before the emergence of any SFPs.
 'E-Journals - Exchange of Experience Meeting', 26 February 1993, The Royal Society, London (organised by the British Library Research and Development Department).
 Invited talk to the staff of CAB-International (CAB-I) on 5 November 1993
 Workshop at the National Centre for Software Technology in Bombay, India, on 6 December 1994.
 Scientific, Technical and Medical.
 For further details see: http://www.njp.org/faq.html
As at January 1999. See http://xxx.lanl.gov/cgi-bin/show_recent_weekdays_graph for latest figures.
 An analysis by Maurice Lines (Lines, 1992)
although carried out for a different purpose agrees with mine in many areas.
 I use 'Web' here as it is the currently available technology - there is nothing implicit in the DJ model that links it to the Web model - any net-based service with a similar functionality could play the role.
 There is, as yet, no consistent terminology in this area - I am using subject gateway to mean a site dedicated to listing links relevant to a particular subject area (e.g. SOSIG for Social Sciences). A gateway may also contain primary information.
 Of course, 'issue' is an anachronism here - first appearance or manifestation my be a better word - but this is a new complex entity and we do not yet have the correct words for its parts, or even have a completely clear description of its parts.
 I use the phrase 'author/producer' to indicate the items on the net are not, and will not, necessarily be in the currently accepted form of the article or book or similar written entity. However, the phase is awkward so for the rest of this article I will use 'author' even though it is understood the more general phrase is intended.
 Selective Dissemination of Information
 Wasteful because there is no need for each user or institution to keep a copy when one can be downloaded on demand. However a few distributed copies for security purposes would be a sensible precaution.
 This model is where part of any income pays for current operation and part is invested for the future.
 One of the main reasons for 'paradigm shifts' according to Kuhn is that the old paradigm (or model) has insoluble problems (anomalies) which often disappear if the new paradigm is adopted.
 The insurance model involves each institution paying a small sum each year, the co-operative model is the formation of a large consortium, e.g., all universities in a country, to provide the service for a membership fee, the pension model is described in the sub-section The hidden roles within the main section The Deconstructed Journal - One possible operational model above.
ACM - Association for Computing Machinery (1995) The ACM Electronic Publishing Plan, Communications of the ACM, April 1995. Available in e-form at http://info.acm.org/pubs/epub_plan.html (last accessed 29/1/99)
Bachrach, S, et al (1998), Who Should Own Scientific Papers? Science, Vol. 281, No. 5382, 4 September 1998, pp. 1459-1460. Available in e-form at http://www.sciencemag.org/cgi/content/full/281/5382/1459 (last accessed 29/1/99)
Bailey, CW, Jr. (1994) Scholarly Electronic Publishing on the Internet and the NII: Charting Possible Futures, Serials Review, Vol 20, No 3, pp 7-16. Available in e-form at: http://info.lib.uh.edu/cwb/schpub.htm (last accessed 29/1/99)
Ginsparg, P (1994) First Steps towards Electronic Research Communication, Computers in Physics, Vol. 8, No. 4, July/August 1994, pp. 390-396.
Ginsparg, P (1996) Winners and Losers in the Global Research Village. Paper presented at the Conference Electronic Publishing in Science, UNESCO HQ, Paris 19-23 February 1996. Available in e-form at: http://xxx.lanl.gov/blurb/pg96unesco.html (last accessed 29/1/99)
Garlock, K, Landis, W. and Piontek S. (1997) Redefining Access to Scholarly Journals: A Progress Report on JSTOR. Serials Review. Vol. 23, No. 1, 1997, pp. 1-8.
Goldhar, J D (1977) Obtaining and Using Information in the Year 2002, IEEE Transactions on Professional Communication Vol PC-20, No 2, September 1977, pp 126-129
Harnad, S (1990) Scholarly Skywriting and the Prepublication Continuum of Scientific Inquiry. Psychological Science Vol 1, pp 342 - 343. Available in e-form at http://cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad90.skywriting.html (last accessed 29/1/99)
Harnad, S (1996) Implementing Peer Review on the Net: Scientific Quality Control in Scholarly Electronic Journals. In: Peek, R. & Newby, G. (Eds.) Scholarly Publication: The Electronic Frontier: MIT Press 1996. Available in e-form at http://cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad96.peer.review.html (last accessed 29/1/99)
Harnad, S (1998) On-Line Journals and Financial Fire-Walls, Nature 395, 10 September 1998. Expanded version available in e-form at http://www.princeton.edu/~harnad/nature.html (last accessed 29/1/99)
Harnad, S. & Hemus, M. (1997) All Or None: No Stable Hybrid or Half-Way Solutions for Launching the Learned Periodical Literature into the PostGutenberg Galaxy. In Butterworth, I. (Ed.) The Impact of Electronic Publishing on the Academic Community. London: Portland Press. Available in e-form at http://cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad97.hybrid.pub.html (last accessed 29/1/99)
IWR (1998) E-journal challenges print pricing spiral, Information World Review, May 1998, p 2
JSTOR (1998) About JSTOR. Available in e-form at http://www.jstor.org/about/ (last accessed 29/1/99)
Kuhn, Thomas S (1970) The Structure of Scientific Revolutions, 2nd edition, University of Chicago Press 1970.
Lines, M B (1992) The Publication and Availability of Scientific and Technical Papers: An Analysis of Requirements and the Suitability of Different Means of Achieving Them, Journal of Documentation, Vol 48, No 2, June 1992, pp 201-219
McKnight, C (1993) Electronic Journals - past, present ... and future?, Aslib Proceedings Vol 45, No 1, January 1993, pp7-10
Odlyzko, A M (1995) Tragic loss or good riddance? The impending demise of traditional scholarly journals, International Journal of Human-Computer Studies, Vol 42, No 1, January 1995, pp 71-122
Odlyzko, A M (1998) The economics of electronic journals. In: Ekman, R and Quand, R (Eds), Technology and Scholarly Communication, University of California Press, 1998. Available in e-form at http://www.research.att.com/~amo/doc/economics.journals.txt (last accessed 28/1/98)
Piternick, A B (1989) Attempts to find Alternatives to the Scientific Journal: A Brief Review, Journal of Academic Librarianship, Vol 15, No 5, 1989, pp260-266
Piternick, A B (1991) Electronic Serials: Realistic or Unrealistic Solution to the Journal "Crisis?", Serials Librarian, Vol 21, No 2/3, 1991, pp15-31
Rowland, F (1994) Electronic Journals, Information UK Outlooks, Issue No 8, June 1994. ISSN 1351 4113
Royal Society (1993) The Scientific, Technical and Medical Information System in the UK, Report published by the Royal Society, The British Library, and the Association of Learned and Professional Society Publishers, 1993. BLR&DD Report 6123, ISBN 0-85403-477-3
Savenije, B (1997) New competitors for the publishers: the cobbler and his last. Electronic Library, Vol 15 No 6, December 1997. pp 431-435.
Schauder, D (1994) Electronic Publishing of Professional Articles: Attitudes of Academics and Implications for the Scholarly Communication Industry, Journal of the American Society for Information Science, Vol 45, No 2, March 1994, pp 73-100
Smith, JWT (1997) The Deconstructed Journal, Proceedings of the ICCC/IFIP Conference on Electronic Publishing '97 'New Models and Opportunities', 14-16 April 1997, University of Kent at Canterbury, UK, Ed: Fytton Rowland and Jack Meadows, ICCC Press 1997, ISBN 1-891365-00-2. pp 73-84. Available in e-form at: http://www.kent.ac.uk/library/ICCC/papers/deconjnl.htm
Sutherland, J (1999), Who owns John Sutherland?, London Review of Books, Vol 21, No 1, 7 January 1999. Available in e-form at http://www.lrb.co.uk/v21/n01/suth2101.htm (last accessed 29/1/99)
Swinnerton-Dyer, Peter (1992) - A system of electronic journals for the United Kingdom, Serials, Vol 5, No 3, November 1992, pp33-5
Walker, T J (1998) Free Internet Access to Traditional Journals, American Scientist, Sept-Oct 1998, Vol. 86, No. 5. Available in e-form at http://www.amsci.org/amsci/articles/98articles/walker.html (last accessed 28/1/99)