Although officially still nothing more than a division of the Stanford University Library, HighWire has already carved out a niche in on-line publishing and established a reputation for excellence that is spreading throughout the scientific world.

"Everywhere we asked, we were pointed toward HighWire," said Kent Andersen, managing editor of Pediatrics, which HighWire just helped put on-line. "They know all the ins and outs of academic publishing."

Barely three years old, HighWire is the brainchild of Michael Keller, who still serves as director of academic information resources at the Stanford Library. "For some time I've been bothered about the treatment of scientific information as a commodity," said Keller, who got his degree in musicology. "Commercial publishers are getting rich by getting information for free from scientists and selling it back to universities. What we're trying to do is provide an economically viable alternative to the nonprofit organizations. We have a certain evangelic fervor to it."

"Our concern was that there wasn't going to be a place for nonprofit scientific journals on the Internet," adds John Sack, associate publisher and director of HighWire and one of the few full-time employees. "We wanted to make sure they would not be crowded out by the profit-making publishers, who have more money to invest."

So far the effort has been remarkably successful. With virtually no start-up capital - except for the vast resources of the Stanford Library system - HighWire has helped launch electronic versions of the Journal of Biological Chemistry, Science magazine, Proceedings of the National Academy of Sciences, Pediatrics, the Journal of Experimental Medicine, and the Journal of Clinical Investigation (the last two published by Rockefeller University Press), and Journal Watch, a bulletin associated with the New England Journal of Medicine.

With only 15 employees, not all full-time, HighWire is still only a line item in the Stanford Library budget. Its main office is a trailer put up on campus after buildings were damaged during the earthquake of 1989.

Although Keller had been thinking about on-line journals for nearly a decade, HighWire didn't take shape until 1993, when Robert Simoni, a tenured professor at Stanford and editor of the Journal of Biological Chemistry, started worrying that the publication was growing unmanageably large. "Although we had switched to weekly publication, we were still at 800 pages and approaching the four-pound limit for second-class postage," he recalls. "We tried a quarterly CD-ROM, but subscribers only complained about the time delay."

One day at a faculty meeting, Simoni asked the networking staff at the Stanford Library if anyone could put the CD-ROM on the university computer system so the JBC could be distributed electronically. After the meeting, Keller approached Simoni and asked if he had considered the World Wide Web. Within a few days, Keller was planning the job.

The biggest hurdle was to convince the board of the American Society for Biochemistry and Molecular Biology. The board took more than a year to make up its mind, yet when it did it demanded the site be up in three months - in time for the Society's January 1995 annual meeting. The prospects were daunting. JBC is the second largest peer-reviewed publication in the world. Other publishers had been preparing on-line journals for more than a year without any results. Yet the librarians accepted the challenge.

"These people approached it the way everyone at Stanford does," said Simoni. "They looked at it as one giant, exciting experiment."

Keller had assembled a team headed by John Sack, former director of the Stanford Data Center, and including Michael Newman, head librarian of the Falconer Biological Library, Sandra Senti, from Stanford Network Services, Ann Mueller, from the Stanford Data Center, Vicky Reich, the library's much-published expert on copyright and intellectual properties, and Mark Weiser, a computer scientist at Xerox Parc, who is also the drummer for Severe Tire Damage, the first rock band to play live on the Internet.

Despite a whole series of unprecedented problems - how to translate Greek characters, mathematical formulas, and other nonstandard characters into the electronic medium using HTML code - the team met the deadline. "We called ourselves HighWire," says Sack, "because we were working the whole time without a net."

Although revenues were $800,000 in 1996 and are expected to double this year, HighWire remains refreshingly informal. "There's a wonderful feeling of getting in on the ground floor of something really special," says Sandy Laws, whose official title is "patrol leader."

Todd McGee, a biology postdoc who is serving as a scientific consultant, says he isn't even being paid yet, although he is being "compensated" in other ways. "They invited me to the first few meetings and haven't told me to stop showing up," he explained.

Yet if there's still a seat-of-the-pants aura around HighWire, you'd never know it from talking to their customers.

"They're absolutely wonderful," says Michelle Hache, products specialist at the Massachusetts Medical Society, which has put its biweekly Journal Watch on-line. "They're easy to work with, they've always got ideas, they never say 'We can't do it.'"

"They're very easy and straightforward," echoes Maria Lebron, business manager of the Proceedings of the National Academy of Sciences. "You ask a question, you get a reply."

The biggest coup to date has been putting the American Association for the Advancement of Science's prestigious Science magazine on-line. "We were already exploring the Internet before we heard about HighWire," said Science editor Ed Rubinstein. "But we were concerned we might overextend our staff. HighWire is not just a vendor, they're a collaborator. They know scientists."

The Science site has proved to be a hit, even with nonscientists. "We're getting 25,000 individual visitors a week, two-thirds of whom had not read the printed magazine," said Rubinstein. "In the past, we pushed information on our readers. Now we're becoming an information interchange."

Publishing on-line is causing many editorial boards to rethink their mission. Most printed journals now publish about 25 to 50 percent of their submissions - sometimes due to peer review but often from lack of space. Even so, costs are growing. Says Todd McGee, the Stanford postdoc: "I remember one person looking at the JBC's four feet of shelf space for a year's issues and saying, 'That's not a journal, it's a database.'"

Some journals, such as the Proceedings of the National Academy of Science and the Journal of Clinical Investigation, have kept the print and on-line versions identical - except for the hypertext and features like direct links to MedLine.

But other journals are beginning to show a divergence. "We have 25 to 35 articles a month in print, but we've added another 10 to 15 on-line," says Kent Andersen of Pediatrics. "We've increased the journal by one-third without adding any paper."

"It's starting to change the nature of the magazine," says Chris Fieldmeier, technology manager at Science. "We're posting data tables and listing results that wouldn't make the printed version. We may eventually print the abstracts of some papers that appear on-line. We're also considering posting all our letters to the editor instead of just selecting a few." The Massachusetts Medical Society's Journal Watch, originally a bulletin of the New England Journal of Medicine, is now publishing on-line semiweekly and in print only semimonthly.

So is anybody making any money out of this? That's a question that usually brings smiles and the same answer. "I don't think you can make money in the early stages," says Maria Lebron of the National Academy of Sciences. "But we're investing in the future."

Many journals are not charging for on-line access, hoping to build a subscription base for the future. "At this point, it's costing us," says Rob O'Donnell, electronic manuscript manager at the Rockefeller University Press. "The Journal of Experimental Medicine is $395 per year but free on-line. Right now we're just kind of feeling our way."

The hope, of course, is that once readers become adjusted to the on-line version, the publishers can start charging. Visitors must now pay for access to Science's site, although it comes free with membership in AAAS. In another approach, full text can be searched and subsequently purchased on a per-article basis for over one hundred journals in the BioMedNet database.

Yet the transition has been slow. "There's a certain institutional reluctance," says Simoni of JBC. "I thought the Ph.D. readership would be more savvy, but from our experience they're still a little behind," adds Chris Fieldmeier of AAAS. "I understand they're focused on their disciplines, but I still find people using Netscape 0.95."

HighWire personnel claim they are still only trying to save money for the Stanford Library. But with offers pouring in and do-it-yourself software in the making, it seems only a matter of time before the line item in the library budget becomes a separate entity. "HighWire will probably grow into something Stanford can't contain," says Michael Newman, one of many employees who is splitting his time. At that point, there could be a corporate spinoff - even an initial public offering.