Will there ever be a science equivalent of Google? Two of the world's biggest science publishing and information firms seem to think that there will. They are about to compete head-to-head to create the most popular tool for searching the scientific literature.

Elsevier, the Amsterdam-based publisher of more than 1,800 journals, has announced that this autumn it will launch Scopus, an online search engine covering abstracts and references from 14,000 scientific journals. Scopus will arrive as a direct competitor for the established Web of Science, owned by Thomson ISI of Philadelphia, the scientific information specialist.

“Scopus will definitely be a threat to ISI,” says one science publishing expert, who asked not to be named. “But ISI will not just let this happen. There will be some kind of arms race in terms of adding new features.”

Many researchers are already wedded to subject-specific databases of scientific information, such as PubMed, for biomedical research. But Web of Science is currently the only service to cover the full spectrum of scientific disciplines and publications. It can also generate the citation statistics that are sometimes used to measure the quality of journals and individual papers.

ISI, which is widely used by libraries worldwide, may be hard to displace. It covers fewer than 9,000 journals, but it has been available in its present form since 1997 and includes a 60-year archive of papers. Thomson ISI says it will extend this to 105 years by the end of 2005. The company also owns the only extensive database on patent abstracts.

Elsevier cannot hope to match this coverage in the short term. The company has been able to draw on its experience of running biomedical and pharmaceutical databases, and developers began compiling a multidisciplinary index two years ago. Even so, when it launches, Scopus will index only five years of references for some journals, rising to ten years during 2005. Data on abstracts will go back further, in some cases to the mid-1960s.

Because Scopus has been built from scratch, Elsevier has been able to work with librarians to develop an alternative to the Web of Science interface, which has been criticized by some users. “Users are very happy with Scopus,” says Steven Gheyselinck, a librarian at the University of Lausanne in Switzerland who has been testing it.

Although Scopus and Web of Science are the only products aiming to cover all of science, other search engines are also under development. The Google of science could end up being Google itself: the company has collaborated with nine publishers, including Nature Publishing Group, to create an engine called CrossRef Search.

This service, a pilot of which appeared last month, allows users to search digital versions of all papers held by the publishers involved and returns links to articles on their websites. Unlike Web of Science and Scopus, which scan through the titles and abstracts of articles, CrossRef Search also searches the full text of papers. Many of the other 300 or so members of CrossRef — a publishers' collaboration established to allow easier linking between citations — are likely to join the service if the pilot is successful.