Without doubt, electronic publications in general, and electronic journals in particular, provide some advantages compared to paper-based documents. Electronic documents typically are delivered with a powerful mechanism to search easily through them. Navigation tools allow readers to jump to particular sections, e.g., to references or graphics, and from there to the relevant section in the body of the text. ``Forward referencing'' provides links to articles which were published later, but cite the original article. Corrections can be included without difficulty. References can be linked to abstracting services, from where abstracts or full texts of cited papers can be obtained, and similar publications can be retrieved based on the original article. Electronic versions of journals usually are available in advance of printed versions, and in addition users can browse the contents tables of forthcoming issues. Journal issues need not be shipped, thus avoiding delays due to mailing systems. Electronic documents can be accessed from anywhere at any time and by as many simultaneous users as needed.
New tools will go beyond these features soon. Graphics will be available together with the underlying data, allowing readers to modify them according to their needs; video or sound sequences as well as computer programs can be embedded; journal issues can be linked to data archives, image libraries, laboratory measurements, or software collections (Boyce, 1996), or users may be able to launch demonstrations or other virtual experiments.
On the other hand, electronic publications, at least if used with today's technology, are bad for browsing and nearly unacceptable for reading online. Interesting documents are likely to be printed, probably even several times, at the users' sites, thus shifting printing costs from the publishers to the scientists' institutions. Further costs arise because special tools, both hardware and software, may be required for downloading, reading, and using electronic publications. Unless a text-only version of electronic publications is provided, terminals need to have graphical capabilities. Specific versions of Web browsers need to be installed in order to correctly display the documents. Appropriate software might be necessary to store or print electronic publications. Bandwidth is unlikely to keep up with demand at least in the near future (Butler, 1996); if network capacities are not sufficiently high, access to documents can be painfully slow or impossible. Even with powerful connections, loading of individual documents on the Web can be extremely time-intensive, for instance if the providers include extensive graphics. Many network users try to work around this problem by simply switching off the display of graphics in their Web browsers.
Costs for electronic publications are not limited to purchasing equipment. The myth of all information on the Internet being free-of-charge has not yet been destroyed. In the first days of the net, many documents were maintained by individuals who spent their private time and personal efforts into setting up network resources. This voluntary attitude has led to the widespread, but wrong assumption that electronically delivered information doesn't cost anything. The truth is that the Internet is just the communication channel; valuable content must be paid for as much as before. Commercial publishers are about to leave their experimental phase during which access to their publications often was allowed without any charge, for testing purposes, and now are establishing subscription rates. Combined subscriptions for print copy plus access to the electronic version of journals typically are more expensive than print-only subscriptions. While the reasons for price increases may be understood by librarians, it can be very difficult to explain them to those who decide on the size of library budgets. As a consequence, we have to fight even harder for our budgets now.