Over the summer, University administrators were trying to perform what seems like a simple, one step task to distinguish between faculty members who have tenure and those who don't so that they could input a change in fringe benefits.

It wasn't quite so easy.

First, the tolks in Holyoke Center had to produce a list of which faculty were tenured and which weren't. Next, they had to get a hold of the thousands of faculty personnel files. Then, after entering the data, Holyoke Center had to produce several sheets of paper, take them to the central administration and re enter the data manually into the system.

In all, the project required 60 hours of work. And that's not fast enough in the 1990s, says Vice President for Finance Allen J. Proctor '74.

Indeed, the University's data system isn't one for the modern age at all. It's 30 years old and "heavily tweaked," Proctor says. Beyond that, acknowledges Deputy Administrator Anne H. Margulies, the system is "inconsistent, sometimes it's inaccurate and sometimes it's difficult to get the answers to relatively simple questions."

But with its ongoing Administrative Data Project (ADP), the University is attempting to leap into the modern era.

The ambitious project, whose final results may not be seen until the beginning of the next century, is designed to improve the quality and consistency of information, while enhancing the efficiency of transactions.

The examination of the University's data system started a new more than two years ago as part of President Neil I. Rudenstine's University-wide academic planning process.

Central administrators quickly realized that "a lot of information they felt was necessary in order to be effective leaders University wide wasn't available in the form that was helpful accurate, timely," Proctor says.

Administrative services such as issuing and collecting checks and hiring processes require numerous steps because they're done manually, according to Lawrence R. Ladd, the University's director of budget and financial planning.

Meanwhile, such basic information as how many people are on a given payroll and how many faculty are tenured or untenured is difficult to retrieve without a mountain of paper Proctor says Even the highest ranking of Harvard's deans have to slog through these processes.

Margulies who is spearheading the ADP. points to personal investigators who don't want to overspend a particular grant account She says a printout of how much money is in a particular account is only available from the University once each month.

As a result, several departments have set up their own data systems, but the duplicate entry leaves too much room for accounting or data-entry errors, Margulies says.

The vast differences in accounting practices across the fiercely decentralized University complicate things further. In fact, information is so hard to centralize and collect that bulk purchasing is greatly complicated, costing the University thousands of dollars annually, says Vice President for Administration Sally H. Zeckhauser, who is on the ADP's executive committee.

As part of its preliminary efforts, the ADP put together a pilot desktop system last December, January and February. The program allowed lab directors and departmental administrators, for example, to make hiring and purchasing decisions via computer, Proctor says.

The program was successful, administrators say, but identified two further problems.

The first is the need to replace the University's core financial systems--its general ledger and accounts payable systems, for example. Proctor says these endeavors "move you from the $10 million range to the $30-$40 million range, which means it's a much higher hurdle in terms of payoff viability."

The second problem was the need for a University network infrastructure, which would be a movement against tradition. Now, the ADP is undertaking the mammoth task of achieving standardization throughout the University.

Right now, the ADP is focusing largely on those two problems, Proctor says.

Administrators are reluctant to predict a date for when a new system will be implemented.

Proctor says the soonest implementation could begin would be July, 1997, which would mean a completion date around the year 2000.

Administrators are also reluctant to peg a cost on the project, but they say that it will largely pay for itself. "We know that if we simplify our work, eliminate duplication, provide people with information they need,... we'll hopefully save money at the same time," Martgulies says, "What started out as a project to fix the data...has evolved into a project into making our work more efficient and simpler and streamlined."

Jonathan N. Axelrod contributed to this report.