ELVYN: The Delivery of an Electronic Version

[Paper given at the JUGL 94 Conference, Strathclyde University, 22 June 1994]

ELVYN: The Delivery of an Electronic Version of a Journal from the Publisher to Libraries

Hazel Woodward, The Pilkington Library, Loughborough University of Technology,

and Fytton Rowland, Department of Information and Library Studies, Loughborough University of Technology

Abstract

The ELVYN project investigated the scenario of dual (print and electronic) publishing of journals in the physical sciences by delivery of the electronic file from the publisher to academic libraries, each of which mounts the file at its own site. The journal chosen for the project is Modelling and Simulation in Materials Science and Engineering (MSMSE), published by Institute of Physics Publishing (IoPP) in printed form.

The stages of the project were: (1) investigation of the requirements of materials scientists and libraries at each of seven sites, chosen for their expected level of interest in the subject matter of the journal; (2) design and implementation of a system for provision of the full file (text and graphics) of MSMSE at each of the sites; (3) recruitment of users and collection of usage and costs data at each of the sites.

The publishers' intention was not to dictate to sites what form of the database they should have, but to seek to meet the different requirements of each site. Thus several quite different implementations were developed, requiring different formats to be delivered by IoPP and delivering different formats to users.

This paper reports stages 1 and 2 and provides preliminary usage data from three of the participating sites, and presents some interim conclusions.

Background

For many years there have been predictions that printed scholarly research journals would eventually be replaced by publications distributed electronically, either online or on some portable medium such as diskettes or CD-ROMs. The reasons for these predictions were, amongst others, cost and speed (Harnad, 1992). The price of scholarly journals has consistently risen faster than both the general rate of inflation and the increase in library budgets (Woodward & Pilling, 1993). Many printed journals have unacceptable publication delays: six months or more in the sciences, often years in the humanities.

Electronic publishing came first in the abstracts and indexes area, where the journals have appeared in parallel printed and electronic forms for many years. These journals, however, rarely contained graphics and tabular matter. The concept of electronic publication of scholarly primary journals was established by John Senders in 1977 and has been the subject of several major research studies during the subsequent seventeen years (Senders, 1977; Mastroddi, 1987; Williamson, 1988; Tuck et al., 1990; Shackel, 1991).

Progress on electronic journals was impeded for many years by the inadequacy of the networks, hardware and software available to most academics. Outside the hard sciences, few people in British higher education had ready access to the Arpanet (predecessor of the Internet), and even the UK national academic network, JANET, was initially used mainly by scientists to access major computing facilities for research number-crunching. Online searching of the major abstracting and indexing databases required use of the packet-switching system to reach hosts like Dialog, and the charges imposed by both the host and British Telecom resulted in limited use by academics, who generally lacked resources to pay for information services. For years the UK Research Councils would not permit research grant moneys to be spent on information services. This situation has changed only very recently. Only in the last five years have all departments in all universities been connected to JANET -- indeed, a few departments in the "new" universities may still not be connected. A current research project in a well-resourced "old" university, Sheffield, shows that, in the early months of 1994, while 81% of academics there used some electronic information service and 51% could access these from their own offices, only 18% used the Internet and only 4% commercial information services such as Dialog. The major services used were, not surprisingly, the university's own OPAC and electronic mail (R. Duffin & G. Sobczyk, personal communication; Anonymous, 1994).

There are commercial services in existence, such as Adonis and those offerred by UMI, which provide journal articles on optical disks for document delivery purposes; however, the only searchable elements are the bibliographical records and abstracts, the main files being page images, and therefore dependent on the existence of a printed journal to put under a scanner. These are not electronic journals in the full sense: that is, they could not be delivered if the printed version did not exist.

There is considerable controversy at present about the model of electronic scholarly publishing that that may prevail. Those models advocated include: a UK national archiving system administered by The Royal Society (Swinnerton-Dyer, 1992); direct electronic publishing by authors of their own material (Harnad, 1994); electronic preprints follwed perhaps by later conventional publishing (Ginsparg, 1994); publishing by the university library of the author's university (Quinn, 1994); publication by the journal's editor on the Internet, free of charge (Odlyzko, 1995); electronic archiving of the full text by the university library, as in Quinn's model, but accompanied by conventional commercial publishing of extended summaries (McMillan, 1994); publication of paid-for electronic-only journals on the Internet by conventional publishers (Fisher, 1994); and dual -- electronic and printed -- publication of existing journals by their present publishers, as in the current project and a number of others listed below.

In the last couple of years, many new electronic periodicals have been established on the Internet (Okerson, 1994). Only a minority of these are scholarly journals in the true sense, reporting full and novel research results, supported by literature references and subjected to rigorous peer review. Nonetheless there are now estimated to be 45 such journals (Woodward & McKnight, 1994) -- about 10% of the total number of electronic periodicals -- and the number is growing rapidly. These new journals are mounted on servers by their academic editors, and information is retrieved from them by academic users without the intermediacy of either a publisher or a library. Almost all of these at present, however, consist of ASCII text only and are distributed by such means as e-mail, gopher and ftp. Subject areas that are dependent on high-quality graphics, especially half-tones, on mathematical symbols and on extensive tabular data tend not to be represented among these new electronic journals. As long ago as 1983 it was clear that these would be the technically difficult areas (Rowland, 1988). In the mathematics and physics fields, TEX has become a lingua franca and facilitates electronic publication (Ginsparg, 1994; Odlyzko, 1995) but it is unlikely to become so for other subjects. World Wide Web and Mosaic facilitate the communication of graphics over networks (Knight & McKnight, 1994), but currently the inclusion of graphics causes unacceptably slow delivery of pages for real-time use. The advent of SuperJANET and NREN should improve this situation but it is not clear how soon all the links -- local, national and international -- in the networks will reach this standard.

At present a number of electronic publishing projects involving commercial publishers (including not-for-profit publishers such as learned societies and university presses) are in progress. These include Red Sage (Springer and the University of California; Badger & Wallace, 1994), Tulip (Elsevier and various American university libraries; TULIP, 1992), Muse (Johns Hopkins University Press and Johns Hopkins University Library; Kelley & Lewis, 1994), the Online Journal of Clinical Trials (OCLC and the American Association for the Advancement of Science; Wilson, 1992); CAJUN (the journals Electronic Publishing and Optical Quantum Electronics); CORE (the American Chemical Society and Cornell University Library; Landauer et al., 1992) and the IEE's Electronics Letters (Pache, 1994). Many of these projects were discussed in detail by Rowland (1993). Typically, they entail the publisher making the files available in a format of the publisher's choice, and in many cases a particular software tool is required by the user to be able to view and print a document at a local terminal. Furthermore, the technologies used differ between the experiments and thus, should they all become routine operations, academic libraries and their users would need different software tools to access the electronic publications of different publishers.

Most of the major publishers of scholarly journals are considering what their policies should be towards electronic publishing of their existing stables of established journals, particularly bearing in mind their need for continuing commercial viability; even the not-for-profit ones also have to be "not-for-loss". So should they go for dual publication or even electronic-only publication? It was against this background that the Elvyn project was planned. It was distinctive in that the publisher sought to discover, before launching the electronic version, what the needs and wishes of potential users, librarians and computer centre staff at various universities were. The intention was not to offer a single product of the publisher's design, but to seek to supply each customer library with a format suiting its own particular needs. Each library in turn would then seek to provide a local online service tailored to the expressed needs of its local potential users. This approach would enable the project team to make a comparative evaluation of the different implementations of the file.

The Elvyn Project

Discussions took place in 1992 between IoPP and the UK's Standing Conference of National and University Libraries (SCONUL) about the possible roles of scholarly publishers and academic libraries in the electronic publishing system of the future. At that time, the first of the new journals distributed free of charge over the Internet were beginning to appear. Learned-society publishers and academic libraries were anxious to continue to be able to provide an effective service for the communication of scholarly communication, and therefore this project was conceived from the beginning as a test of the viability of the specific scenario for electronic publishing that retains both the publisher and the library.

At the time IOPP was planning the launch of a new journal, Modelling and Simulation in Materials Science and Engineering (MSMSE), to appear quarterly commencing in October 1992. Production of this journal was planned from the start to use TEX and to make as much use as possible of authors' own keyboarding of input. Thus it seemed an appropriate vehicle for an electronic publishing experiment: all the data existed in machine-readable form (or, in the case of the graphics, could be scanned to provide TIFF files), and therefore reliance would not need to be placed on page images to provide a display file. Nor was it an easy target: the journal has extensive mathematics and graphics, including line-art, half-tones and colour illustrations that have to be reproduced at high quality if information content is not to be lost.

The market research that had preceded the decision to launch this journal had identified departments that had significant concentrations of researchers in the subject area of the journal.

The project as agreed between SCONUL and IoPP has been carried out from a base at the Department of Information and Library Studies at Loughborough University of Technology with financial support from British Library Research and Development Department. Seven of the sites known to have a concentration of potential users of MSMSE were selected as test sites and their Librarians agreed to take part. They were also chosen in part to reflect some diversity: a non-academic library (the Harwell Laboratory), a new university (the University of Hertfordshire), a departmental library (the Cavendish Laboratory in the University of Cambridge), a university library in a technological university where there are no departmental libraries (Loughborough University), a technological college with strong departmental libraries (Imperial College), a large general college in the University of London (University College), and a fairly small provincial university (University of Exeter) where, however, there was already established research activity about Standard Generalised Markup Language (SGML).

The underlying concept of the project was that IoPP would not offer a preconceived product to the sites, but would seek to provide each site with what it required based on its individual circumstances: the hardware, software and networks available there, and the views of its materials science researchers and its librarians about the kind of service that was desirable.

The project therefore fell into three phases: a predelivery phase during which interviews were held with librarians and potential users at the site; a prelaunch phase during which systems were designed and set up at each site and the MSMSE files were provided by IoPP in the format requested by the particular site; and a running phase during which the MSMSE files were available on the campus network at each site and users could access it and provide usage feedback. In addition to usage data, costs data were also collected from both the publisher and the libraries, in order that assessments might be made of the financial implications for both parties of a possible widespread swing to dual-version journals.

The Predelivery Phase

A questionnaire was devised, containing sections for materials scientists and for librarians, and including both factual and attitudinal questions. It was administered by an interview-questionnaire technique by one of us (F.R.) and a consultant, Peter Such, during a series of visits to the seven chosen sites in the autumn of 1992.

The questionnaire obtained information from the libraries about the IT infrastructure available within the university: networks, machines available in the library and in users' offices, and standard software available on the network or widely adopted among potential users. Information was also sought about the administrative and managerial stuctures both of the library and of the university as a whole, especially as regards the relationship between the library and the computer centre.

Potential users (academic materials scientists) were asked during the winter of 1992-93 about their own use of computers for information tasks, their patterns of use of the library, and their opinions about possible electronic journals.

The results of the interviews were analysed by Dr David Pullinger of IoPP, using a deconstruction technique. This analysis arrived at a conclusion, for each site, about what features of the MSMSE service would be essential at that particular site to satisfy the expressed needs of the potential users within the organisational and infrastructural constraints described by the librarians. At this stage, early in 1993, it was assumed that what had emerged from this deconstruction analysis would be the basis for the service provided at each site, and thus for what form of files IoPP would have to provide to each site as input. Full details of this phase of the project are given in the final report on the project (Meadows & Rowland, 1994).

The Prelaunch Phase

In practice, the development process at each site was iterative. The sites varied in the extent to which library staff could implement the chosen system themselves. In most cases, it was necessary to involve a contact person in computing services as well as a contact person in the library. Precisely what system was provided at the site then emerged from discussion between these two people and, in most cases, also with technical staff at IoPP. It became clear that the technical complexity of mounting files containing graphics and mathematics as well as text, and of seeking to provide output (both on-screen and as printouts) that approached the typographical quality of printed journals, had been seriously underestimated. Furthermore, with the rapid growth in student numbers that was taking place at that time, both library and computing services staff in universities were under severe pressure of work. For all these reasons the prelaunch phase fell behind schedule and it was not until the Spring of 1994 that most of the sites were essentially ready to provide a service. The actual list of sites eventually participating, and the systems they were using, differed from those planned early in 1993.

For various (largely organisational) reasons it became clear that neither the Cavendish Laboratory not Imperial College would be able to participate. At a quite late stage, therefore, two other sites with substantial concentrations of materials scientists with relevant research interests were added to replace them: the University of Oxford (the Radcliffe Science Library) and the University of Manchester/UMIST joint Materials Science Centre. In addition, the library of a technological university in Sweden, Chalmers University, Gothenburg, was interested in the project and came in as an additional site. To enable these new sites to mount the files quickly, however, they were provided with the system that had been developed at Loughborough. Thus this system was eventually used in four places, while the other four -- Harwell, Hertfordshire, University College and Exeter -- each adopted an individual approach. Appendix 1 briefly describes the technical approach finally adopted at each site.

The Loughborough system was described in detail at the ELVIRA conference in May 1994 (Knight & McKnight, 1994, 1995). At this site it was decided, in the autumn of 1993, to use World Wide Web and Mosaic. Developments around the world since that date have shown that this was a good choice, since these tools are rapidly acquiring the status of a de facto standard, and by their use it may reasonably be hoped that most potential users of an electronic journal will be able to access it without the need for local purchase of more specialised tools.

Costs

The test sites, and IOPP, were asked to provide data on the costs incurred during the prelaunch phase. These consisted mainly of staff time, and at those of the test sites that have so far supplied these data the staff costs were of the order of 3000 pounds.

One or two sites also had to enhance hardware for the project, which also incurred costs; at Hertfordshire these amounted to 2750 pounds.

An attempt was also made to collect comparative cost data on the costs of ordering, accessioning and shelving a new quarterly printed journal, but only Loughborough was able to produce some data on this point. A study carried out in Loughborough during the 1980s (Wilson, MacDougall & Woodward, 1986) was therefore updated and extended to provide this information. Based on a time of 3.4 minutes for a library assistant to process one issue of a journal the cost for handling a single issue is 0.37 pounds (i.e. 1.48 pounds per year for a quarterly journal). This applies to each issue; there is also a start-up cost due to the need to catalogue the journal when it first appears. Such cataloguing takes approximately 15 minutes, which gives a cost per journal title of 2.70 pounds.

At I0PP, the costs were greater; 11500 pounds were expended by IOPP on the project in this phase.

The Running Phase

Usage data: students

Three sites have provided some usage data already: Loughborough, Oxford and Chalmers. All three of these use the Loughborough system, so we are not yet in a position to make comparisons between the systems. Usage data from all of the sites will be collected later, so that further reports on the project will be able to compare user reactions to the different implementations.

At Loughborough, Chalmers, and in the future probably at Manchester too, a questionnaire composed by one of us (H.W.) has been used to test out students' usage of the electronic version of MSMSE. The questionnaire (Appendix 2) set out five tasks, each one using a different part of the system, and asked users whether they found the task easy, slightly difficult or very difficult. They also had scope to make comments about each part of the process. Question 1 required browsing from the contents lists; question 2, accessing graphics; question 3, text searching through the complete files; question 4, text searching through a specific paper found by browsing; and question 5, accessing the reference list of a paper by using the hypertext link from the main text. Figure 1 shows the results of the use of this questionnaire with three groups: ten materials science students at Loughborough, eleven information studies students in Loughborough, and eleven information studies students at Chalmers. The Manchester students will be materials scientists. All of these students were on taught masters courses.

All the information studies students found question 1 (browsing mode) easy, but half the materials scientists found it slightly difficult.

Accessing the graphics, on the other hand, was more difficult. Two of the Loughborough information studies students failed to find the figure in the paper concerned completely -- the only two males in the group. (One may wonder whether there is any significance in that.) Of the other nine, one found it easy, five slightly difficult and three very difficult. At Chalmers, the information studies students found it easier to access the figure: six said it was easy and four slightly difficult, with one person not grading it at all. Of the materials scientists at Loughborough, two said it was easy, four slightly difficult and four very difficult.

For technical reasons, Chalmers students could not attempt question 3, text searching. At Loughborough, information students rated it easy (2), slightly difficult (8) and very difficult (1), while materials science students rated it easy (4), slightly difficult (5) and very difficult (1)-- no great difference between the disciplines there.

Again because of technical problems, the materials science students at Loughborough could not attempt questions 4 & 5. Information studies students at Loughborough rated question 4 (text search within a document) as easy (just one person dissenting with "slightly difficult"), while at Chalmers six found it easy, one slightly difficult, one did not grade it, and three for some reason did not attempt the question. Perhaps this was due to language difficulties since these three students wrote their comments in Swedish.

Question 5 -- jumping to the references -- was found by Loughborough students to be easy (8), slightly difficult (2) and very difficult (1), while Chalmers students found it easy (5), slightly difficult (1), very difficult (2) and not graded (2), while one person did not attempt the question.

If we analyse the comments made by the students, the commonest remarks were: that the system was slow; and that finding graphics was difficult because one had to read through the text to find the reference to the figure in the text, and then click on that reference to it in the text to bring up the graphic itself. In addition, in making the hypertext jump from the reference point in the text to the reference list, the link did not go the specific reference but to the top of the reference list, from which one has to scroll down.

Usage data: staff

Four materials science staff at Oxford and three at Loughborough have so far responded. At Oxford a form was filled in, which among other things asked users to rate speed of access, image quality, text quality and ease of manipulation on a scale from very good to very poor. One of the four users rated it very poor on all counts (because he failed to get access to it at all!), but the other three rated it good to very good. All three of these users had, however, used it only in browsing mode, and not in search mode, and only one of them had accessed images or attempted to print anything out.

At Loughborough three staff users so far have been interviewed by one of us (H.W.). All of them liked the ability to use the journal at their own desk, and they had used it several times for half an hour or so at a time. They were critical of the slow access speed, the splitting up of articles into sections (which was done to compensate for the slow speed) and the fact that figures have to be displayed separately from the text. The hypertext link to the reference list was rated positively.

Running costs

The ongoing costs at each site associated with the accession of each new issue of MSMSE are currently being calculated.

At IoPP, on the other hand, running costs per issue have already been estimated, since several issues have been converted into the various formats requested by the sites already. A figure approaching 4000 pounds per issue has been arrived at; again, this consists mostly of staff time.

Interim conclusions

We have been able to arrive at certain tentative interim conclusions.

Much has been learned at the publishers as regards conversion, development, administration and production presentation methods, but no absolute answers have emerged. There has been a steep learning curve for the publisher's staff.

To arrive at a consensus among a group of potential users at each site and library staff who would have to implement the system was not easy, and in some cases impractical choices emerged from the first stage and had to be revised.

The implementation of the system at each site was a larger technical task than had been envisaged, almost always requiring active involvement of computing staff as well as library staff, and took a longer time than expected.

Not all of the chosen sites where concentrations of materials scientists exist were able to implement the project. Thus, the technological sophistication of a site in terms of its research laboratories is not necessarily matched by adequate interest, ability, or administrative structures in the library to render this type of publication practicable.

In spite of the expected level of interest at the chosen sites, recruitment of users has been difficult. This may be due to people's unwillingness to learn a new system for the sake of a single quarterly journal, or may reflect a lack of interest in electronic journals generally among research scientists.

Only very tentative overall conclusions can be drawn at this stage. A single quarterly journal has been considered, and while different implementations have been put in place, the limited usage and costs data available at present represent only one of the implementations, that using World Wide Web and Mosaic. However, this implementation is, arguably, the best choice, in view of their increasingly widespread availability in the academic world. Nor is it clear that the model chosen for this experiment (one in which both publishers and libraries retain a role) will prevail. It has been demonstrated through this research project that both these parties can learn rapidly how to operate an electronic journal and can continue to add value to information. Questions remain about the economic viability of scholarly publishers in this new electronic world and controversy continues on this point (Harnad, 1994; Odzylko, 1995).

References

Anonymous (1994) Information skills for student centred learning at Sheffield. CTILIS Newsletter, 5(3), 2 (news item written by F. Rowland).

Badger, R. & Wallace, M. (1993) Electronic journals: The Red Sage Approach. Newsletter on Serials Pricing Issues, no. 91, paper 2.

Fisher, J. H. (1994) Editorial costs. Message posted on the Publishing E-

Journals: Publishing, Archiving and Access discussion list (VPIEJ-

L), 29 July 1994.

Ginsparg, P. (1994) First steps towards electronic research communication. Computers in Physics, 8(4), 390-396; http:// xxx.lanl.gov/blurb/.

Harnad, S. (1992) Interactive publication: extending the American Physical Society's Discipline-Specific Model for Electronic Publishing. Serials Review (special issue on Economic Models for Networked Information), 18 (1/2), 58-61.

ftp://princeton.edu/pub/harnad/Harnad/harnad92.interactivpub

Harnad, S. (1994) Implementing Peer Review on the Net: Scientific quality control in scholarly electronic journals. In Proceedings of an International Conference on RefereedE lectronic Journals: Towards a Consortium for Networked Publications. University of Manitoba, Winnipeg, Manitoba,Canada, 1-

2 October 1993 (in the press).

Kelley, T. & Lewis, S. Project Muse: A new venture in electronic scholarly communication. Newsletter on Serials Pricing Issues, no. 109, paper 2.

Knight, J.P. & McKnight, C. (1994) Delivery and use of an electronic version of a journal. Paper presented at the ELVIRA Conference, Milton Keynes, UK, 3-5 May 1994.

Knight, J.P. & McKnight, C. (1995) Project Elvyn: Implemeneting an electronic journal. Computer Communications, in press.

Landauer, T., Egan, D., Remde, J., Lesk, M., Lochbaum, C. & Ketchum, D. (1992) Enhancing the usability of text through computer delivery and formative evaluation: the SuperBook project. In Hypertext: A Psychological Perspective (McKnight, C., Dillon, A. & Richardson, J., eds). Chichester, UK: Ellis Horwood.

McMillan, G. (1994) VT (Virginia Tech) Model. Message posted on the Publishing E-

Journals: Publishing, Archiving and Access (VPIEJ-L) discussion list, 27 July 1994.

Mastroddi, F. (ed.) (1987) Electronic Publishing: The New Way to Communicate. Proceedings of a Symposium held in Luxembourg, 5-7 November 1986. London: Kogan Page, for the Commission of the European Communities.

Meadows, J. & Rowland, F. Delivery of an Electronic Version of a Journal from Publishers to Libraries. British Library Research & Development Department report, in the press.

Odlyzko, A.M. (1995) Tragic loss or good riddance? The impending demise of traditional scholarly journals. International Journal of Human-Computer Studies, in the press; condensed version in Notices of the American Mathematical Society, January 1995; ftp://netlib.att.com/netlib/att/math/odlyzko/tragic.loss.Z

Okerson, A. (1994) Directory of Electronic Journals, Newsletters and Discussion Lists, 4th edn. Washington, D.C.: Association of Research Libraries.

Pache, J. (1994) Electronics Letters online. Paper presented to the Internet World International Conference, London, UK, 11 May 1994.

Quinn, F. (1994) A role for libraries in electronic publication. EJournal, 4(2), lines 68-

416.

Rowland, F. (1993) Electronic delivery of research information. Serials, 6(3), 40-50.

Rowland, J.F.B. (1988) Electronic journals in chemistry -- a project in the DOCDEL programme. Chapter 6 in Electronic Publishing, State of the Art Report 15:6, ed. by Williamson, R. Maidenhead, UK: Pergamon-Infotech, pp. 75-82.

Senders, J. W. (1977) An online scientific journal. Information Scientist, 11(1), 3-9.

Shackel, B. (1991) BLEND-9: Overview and Appraisal. British Library Research Paper no. 82. London, UK: The British Library.

Swinnerton-Dyer, Sir H.P.F. (1992) A system of electronic journals for the UK. Serials, 5(3), 33-35.

Tuck, B., McKnight, C., Hayet, M. & Archer, D. (1990). Project Quartet. LIR Report no. 76. London, UK: The British Library.

TULIP (1992) TULIP: An Irregular Update, no. 1. New York, NY: Elsevier Science Publishers.

Williamson, R. (ed.) (1988) Electronic Publishing, State of the Art Report 15:6. Maidenhead, UK: Pergamon-Infotech.

Wilson, D.L. (1992) Major scholarly publisher to test electronic transmission of articles. The Chronicle of Higher Education, 3 June, A17 and A20.

Wilson, J.M., MacDougall, A.F. & Woodward, H.M. (1986) Economic consequences of libraries acquiring electronic journal articles. British Journal of Academic Librarianship, 1(3), 228-235

Woodward, H. & McKnight, C. (1994) Electronic journals: issues of access and bibliographic control. Serials Review, in press.

Woodward, H. & Pilling, S. (1993) The international serials industry: An overview. Chapter 1 in The International Serials Industry (Woodward, H. & Pilling, S., eds). Aldershot, UK: Gower Publishing, pp. 1-22.

Appendix 1. Technical solutions adopted at the sites

Manchester, Oxford and Chalmers use the Loughborough system.

Loughborough University of Technology

SGML is taken from IOPP and converted to HTML using the Copenhagen SGML Tool (CoST); each paper is converted in about 15 minutes. HTML is mounted on a World Wide Web server and viewed using X Mosaic. The mathematics and other non-ASCII characters within the text are taken from IOPP as TEX embedded in the SGML; a complex series of software conversions converts these into X bitmaps which are displayed at the correct inline position within the HTML files. Sections of papers (abstract, introduction, materials and methods, etc.) are mounted as separate files in order to speed up display. For bibliographic references, a hypertext link from the reference point in the text takes one to the reference list.

Graphics are taken from IOPP as TIFF files; these are displayed using External Viewer (xv), with links from the Figure captions in the HTML files that spawn xv and display the graphic.

Tables originally had to be handled manually but later, after consultation with IOPP staff, it became possible to convert them from SGML to HTML automatically.

Browsing is accomplished by entering from a main menu listing the available issues of MSMSE; clicking on an issue brings up the contents list of the issue. Clicking on the title of a paper in the contents list brings up a list of sections of the paper. Clicking on one of these brings it up, including any inline graphics for the mathematics. From here, clicking on the caption brings up any graphic or table cited in the section.

As an alternative to browsing, a small search engine written in a mixture of TCL and csh is also provided. This can be initiated from a button on the main menu.

University College London

TEX is taken from IOPP and is used to generate a full-text index in XWAISQ, a client-server system that provides for full-text indexing. A search using keywords will rank the hits according to the number of occurrences of the search terms within the documents, and from this ranked list links are made to the documents themselves for display. At this stage a choice between "journal" (text) and "figures" is offered, so the user can see the figures first if preferred. PostScript files are also taken from IOPP and are used for the display, using GhostView as the viewing tool. The entire text of each paper (including tables) is included in one PostScript file, while each Figure is held as a separate encapsulated PostScript file; this choice was made because graphics are much slower to display. one can have two windows open, one for text and the other for figures.

University of Hertfordshire

The files are taken from IOPP in TEX and an ASCII file is generated from this. The ASCII text is used to provide a full-text searching facility using WAIS. In addition, known users are included in an e-mail list and receive the table of contents and the abstracts of the papers by e-mail. Having identified a hit by either of these means, users then obtain the display files, which are held in LaTEX. The incoming TEX file is therefore also converted into LaTEX to facilitate display and printing out of the full papers including graphics. Some users only have alphanumeric connections, so they cannot display the hits, but they can place an order online for prints from the library. Those users who are equipped to do so can display the documents an their screen and/or print them out at their own office from the LaTEX files.

Harwell Laboratory

TEX is taken from IOPP and an ASCII file of the full text is generated from it. The full text is mounted on Harwell's own search system, STATUS, on a DEC VAX machine, and is therefore searchable from anywhere on the site. This search file does not contain the mathematics or the tables. For printout, the user has a choice between TEX and PostScript, either of which can be downloaded over the site network from the Library machine to the user's local machine; the user's choice will depend on what local software is available for interpreting the files in that particular user's laboratory. If none is available, it is possible for the user to order prints to be made in the Library and to be sent to the user in hard copy through the internal mail.

University of Exeter

This site is not yet operational, owing to delays in the delivery of the software that they have decided to use. They take SGML input from IOPP, and will be using DynaText, a commercial viewer from Database Publishers Ltd, for accessing it. Graphics are taken from IOPP as TIFF files. We do not yet know precisely how easy it will be for the conversion to DynaText to be made, or what the user interface will look like, as we in the project team have not seen DynaText ourselves.

Appendix 2. Questionnaire used with student users

ELECTRONIC JOURNAL PROJECT. MODELLING AND SIMULATION IN MATERIALS SCIENCE AND ENGINEERING.

Questionnaire for MSc Students in Materials Engineering

Before you start. Please note that throughout the questionnaire there are spaces for you to comment on any thoughts or impressions you have on using the journal. Be assured that we are not testing you on the contents of the articles -

we are interested solely in your reaction to the electronic version.

1. Browse the contents page of volume 1, issue 2. Look at the abstract of the paper by Becker and others.

According to the abstract what do the crystal plasticity models account for?

.............................................................................................................................

..............................................................................................................................

How easy was it to locate the abstract?

[ ] easy [ ] slightly difficult [ ] very difficult

Please describe any problems or impressions connected with this search.

..............................................................................................................................

Now go to the article's section headings.

Any comments? .................................................................................................

..............................................................................................................................

2. From the section headings screen, locate figure 1 and bring it to the screen.

What is the letter by the thick black arrow? ....................................................

How easy was it to find and bring the figure to the screen?

[ ] easy [ ] slightly difficult [ ] very difficult

Please describe any problems or impressions connected with this search

..............................................................................................................................

Now click on the top left hand corner to drop the menu down. Select and click on close

to put away figure.

Any comments? ...................................................................................................

...............................................................................................................................

3. From the introductory screen, perform a keyword search for an article which discusses tessellation. When asked to enter search keyword, move cursor to the grey box.

Any comments? ...................................................................................................

.............................................................................................................................

Which article discusses tessellation?

First author ..........................................................................................................

Article title ..........................................................................................................

How easy was it to flnd the article on tessellation?

[ ] easy [ ] slightly difficult [ ] very difficult

Please describe any problems or impressions connected with this search.

..............................................................................................................................

Now return to introductory screen by clicking on home button.

4. From the introductory screen, go to the Discussion section of the first paper of volume 1, issue 1.

Any comments? ...................................................................................................

...............................................................................................................................

Select file menu from top of screen and click on find in current ...

Move cursor into box. Type in fourier and click on find. This will perform a keyword search within

this section of the article.

What word follows fourier in the text? .............................................................

How easy was it to perform this keyword search?

[ ] easy [ ] slightly difficult [ ] very difficult

Please describe any problems or impressions connected with this search.

..............................................................................................................................

Click on dismiss to close search box.

5. While still in the Discussion section, select reference 21. Give the title of the article.

.................................................................................................................................

How easy was it to find the full reference?

[ ] easy [ ] slightly difficult [ ] very difficult

Please describe any problems or impressions connected with this search.

..............................................................................................................................

Thank you for completing the questionnaire. You may now exit from the system, or if you wish continue browsing.

It would help us if you could print your name below and return the questionnaire to Hazel Woodward in the Pilkington Library.

Name: ..............................................................................................................................

Figure 1. Usage data

Q.1 Q.2 Q.3 Q.4 Q.5

Loughborough Information students

Easy 11 1 2 10 8

Slightly difficult -

5 8 1 2

Very difficult -

3 1 -

Other -- 2 -

Loughborough Materials Science students

Easy 5 2 4 -

Slightly difficult 5 4 5 -

Very difficult -

4 1 -

Other -

- -

Chalmers Information students

Easy 11 6 -

6 5

Slightly difficult -

4 -

1 1

Very difficult -

Other -

1+ -

4$ 3~

Notes: *Failed to answer the question; +Not graded easy or difficult; $One not graded; three did not attempt the question; ~Two not graded; one did not attempt the question