Cambridge University is to create an extensive online archive, whose content and structure will be freely available to any institution that wants to copy it.
Academics will be encouraged to deposit their scholarly papers, theses, teaching and other materials into a central university repository that is based on technology developed by the Massachusetts Institute of Technology.
Peter Fox, Cambridge's university librarian, said: "We have been concerned for some time about the amount of digital material being created in the university, apparently without any provision being made for its long-term preservation and access."
DSpace@Cambridge will be funded for its first two years by £1.7 million from the Cambridge MIT Institute. MIT developed the open-source DSpace repository with Hewlett-Packard laboratories. The software can be downloaded free by anyone.
Cambridge is so far the only UK institution to join the MIT project, which is being piloted by six US universities. In particular, Cambridge will develop the digital preservation applications to feed back into the federation using the software.
The system will be able to manage a range of formats, such as multimedia, interactive teaching programmes and databases. DSpace will be used to capture, index, store, disseminate and preserve materials in digital form.
Peter Morgan, project director at Cambridge's library, said the first phase would be able to accommodate 10 terabytes of data, or about five times as much as an average academic research library. The first content should be available later this year.
Mr Morgan said: "The content will be openly available as long as the creators say it can be. When users place files in the repository, they will have to make the decision whether they want them to be available worldwide."
• Edinburgh University and bibliographic software company ExLibris have won a £1 million contract to set up the first online catalogue of all journals and newspapers in British research institutions, writes Olga Wojtas.
The four-year project will cover databases from more than 200 libraries, including the British Library, smaller specialist libraries and university collections.
Peter Burnhill, director of Edina, the national datacentre at Edinburgh, and project director for the new Suncat catalogue, said that journal records were the "poor cousins" of book catalogues. He added that a single centralised database would save time and trouble for academics, students, librarians and the public.