(FORTUNE Magazine) – LAST MONTH a delegation of high-ranking executives from AT&T Bell Laboratories in Murray Hill, New Jersey, made a quiet pilgrimage to the sunlit Palo Alto, California, headquarters of Hewlett-Packard Laboratories. The object of their quest was not some esoteric technology but a new alchemy. They are studying technology transfer -- the process whereby a business transmutes the discoveries of its scientists into lucrative products. The visit to H-P reflects a dramatic change on the American research scene. A decade ago Bell Labs would hardly seek advice from anyone; instead the world beat a path to Bell Labs' door. America's premier invention factory, it had produced the sound motion picture, the transistor, the solar cell, and the laser, plus epochal discoveries in science, such as evidence that a Big Bang ! created the universe. Technology giants including IBM and Xerox took AT&T as a model in organizing research efforts of their own. For decades the AT&T formula seemed to work: Hire brilliant scientists, turn them loose in search of basic knowledge, and wait for breakthroughs of vast commercial potential. But in recent years the glare of international competition has exposed a crippling flaw in that scheme. It affects not just Bell Labs but all of American industry as well. With a handful of exceptions -- companies like Hewlett-Packard and Du Pont that have mastered the subtle chemistry of technology transfer -- American businesses have proved to be more skilled at making discoveries than at capitalizing on them. U.S. companies last year spent a walloping $68.8 billion on R&D -- more than the R&D budgets of Western European and Japanese companies combined. Yet America's competitive position has eroded alarmingly in many high-tech fields, from the etching of microchips to the building of skyscrapers. With Japanese companies in particular boosting their R&D budgets, technological competition is sure to intensify. America's industrial laboratories are under heavy pressure to couple the work of their scientists and engineers more closely with the demands of the market. Says AT&T astrophysicist Arno Penzias, who shared a Nobel Prize for his work on the Big Bang and now oversees Bell Labs research: ''Five years ago managers here didn't even use the word 'customer.' Now each of us has two jobs: working in corporate research and serving on a team connected to one of AT&T's product areas.'' Technology transfer, if business can learn to practice it efficiently on a broad scale, can help America right its trade imbalance and enhance the nation's wealth. It is precisely the tonic U.S. competitiveness needs. In principle, technology transfer consists of a sequence of simple steps: A research scientist comes up with an insight, which is applied by engineers in a prototype, which in turn serves as the basis for a product that customers buy. In actuality, the process is so difficult to manage that, according to Frank Carrubba, director of Hewlett-Packard Labs, it gives new meaning to ''hit or miss'': ''You're standing there with a handful of darts aiming at a bull's-eye on a suspended dartboard swinging in the wind. You hope to hit the bull's-eye. But you're lucky if you hit the dartboard at all.'' Even the most prestigious R&D operations suffer horrible lapses in technology transfer. Important advances may languish on the laboratory bench while competitors replicate them and get rich; products may emerge poorly conceived, clumsily designed, or ill timed. A classic flop at Texas Instruments involved a revolutionary computer chip known as magnetic bubble memory. TI nurtured the technology for seven years, in hopes of capturing part of the huge market held by ordinary electronic memory chips, known as DRAMs. But the company's engineers underestimated the speed with which the older technology would evolve. By the time magnetic bubbles were ready for the market, DRAMs had become so capable and cheap that computer makers refused to switch. The program cost TI tens of millions of dollars. IF ANY COMPANY holds the key to technology transfer, it is Carrubba's H-P. The $12-billion-a-year maker of computers and instruments maintains a stunningly high rate of innovation: More than 50% of sales derive from products developed within the past three years. By contrast 3M, also one of the nation's top innovators, derives 25% of its revenues from products developed within the past five years. Hewlett-Packard estimates that fully 60% of the research conducted in its labs finds its way into product applications. The phrase drummed into every researcher's head is ''time to market'' -- shorthand for Chief Executive John Young's conviction that products must be brought from laboratory to market quickly and be done right the first time. Under Young's leadership, H-P has devised ambitious new methods for gauging the effectiveness of R&D. The broadest and most important is what Young calls ''breakeven time.'' It is the total elapsed time of a technology transfer, beginning with a scientific investigation and ending only when the profits from a new product offset the cost of its development. Today on average H-P needs three to five years to develop and launch a product, and an additional 18 to 24 months to recoup the cost of development. Young wants to cut that in half by 2000. Technology transfer is in essence a management process that is supremely low tech. Says George Heilmeier, senior vice president at Texas Instruments: ''You delude yourself if you think that the emphasis in technology transfer is on technology. It's a humanistic task, not a technical task.'' Adds Mary Good, a chemist who is senior vice president of technology at Allied-Signal: ''Technology transfer is as much a process of developing contacts, conduits, and advocates for what you do as it is the research.'' Nowhere is this more evident than in the work habits of executives like Carrubba. He believes face-to-face conversation to be so essential to the transfer of ideas that he avoids technology in his daily work. Even though he is a highly accomplished computer scientist at a company renowned for its internal electronic networks, Carrubba uses the telephone only when absolutely necessary and does not have a computer on his desk. (His secretary handles electronic mail addressed to Carrubba on her own PC.) He describes himself as a ''technology broker'' and spends the bulk of his time in meetings that bring his researchers into contact with one another and with the world outside. His most important task, Carrubba says, is to maintain a practical balance between long-term research that can enable H-P to thrive tomorrow and applied R&D that allows it to survive today. Like almost all important technology companies, H-P organizes its R&D in a hub-and-spokes pattern. The hub is the corporate lab, where projects typically aim at achieving practical results three to seven years into the future. The spokes are smaller R&D operations located within corporate divisions, where projects typically have a one-year or two-year horizon and often entail the incremental improvement of existing products. As Carrubba explains, his 1,000-person central lab does ''R&D with a capital R and small d, while the divisions do R&D with a small r and a capital D.'' The 9,000 or so researchers and product developers attached to H-P's divisions report to their own chiefs of R&D, not to the central lab. Each development group is encouraged to concentrate on its division's problems, ''not to sing from the same sheet of music,'' as Carrubba puts it. PRESSED to get more out of their scientists and engineers, some companies lower their sights, cutting back on long-term research or even dispensing with the central lab entirely. Borg-Warner, for example, dismantled its 200-person research center in Des Plaines, Illinois, after a leveraged buyout. Allied- Signal, which spends some $650 million a year for research and engineering, has scaled back the funding of its corporate lab in favor of the labs in its business units. The shift, says R&D director Good, has improved the units' competitiveness, but she is concerned about its effect on Allied-Signal's future: ''Decentralization is very effective for maintaining and growing businesses that you're in. But you lose the ability to expand your company beyond them.'' To avert that risk, companies like H-P keep their central labs intact. This strategy works only if there is sufficient technology transfer that the income from new products justifies the expense of the lab. Alexander MacLachlan, senior vice president of technology at Du Pont, presides over a lab that employs 5,400 researchers in the suburbs of Wilmington, Delaware. The lab is the birthplace of nylon, and MacLachlan asserts that it has given Du Pont an enormous strategic advantage over the years. But he adds that from the moment chemist Wallace Carothers invented the polymer in 1934, Du Pont has grappled with the issue of technology transfer: ''If it had been up to Carothers, we would still be refining nylon rather than selling it. He showed us the basic problem of corporate research -- the tension between people who really have to deliver something to a customer and those who are interested solely in understanding why things work.'' During six decades Du Pont has tried five approaches to technology transfer. In the beginning the central lab was an ivory tower where scientists like Carothers were insulated from commercial considerations. Then came what MacLachlan calls the ''ask for help'' model, in which headquarters encouraged operating divisions with technology needs to petition the central lab for research. Next was a ''control through funding'' scheme that put power in the hands of the divisions: The central lab was required to secure half its operating budget from them. Du Pont scrapped that arrangement after researchers began wasting too much time pitching programs to the divisions. The fourth plan, in use until 1985, was the ''halfway house.'' It split the corporate laboratory staff in two, with one group being responsible for basic research and the other for commercial applications. Paradoxically this slowed the transfer of technology: Isolated from the market, the applications team had little sense of what Du Pont's customers really wanted. Finally Du Pont hit upon the collaborative model it uses today. The company divided its main lab into so-called centers of excellence, each devoted to a particular technology such as advanced ceramics or microbiology. Within each unit, scientists doing basic research work alongside product experts from the divisions. ''The idea is to have these people almost touching each other,'' ! MacLachlan explains. Familiarity breeds good products: One center, for example, recently developed catalysts for producing compounds to replace chlorofluorocarbons, the ozone-destroying coolant used in air conditioners. To promote the intensive exchange of ideas among researchers and engineers at Hewlett-Packard, Frank Carrubba reached back into his childhood. A native of working-class Waterbury, Connecticut, he recalled how people in his neighborhood had congregated on their front stoops and street corners to gossip and trade news. Carrubba is creating a neighborhood atmosphere at H-P as the company's lab buildings are remodeled. Open areas where researchers can meet are being carved out amid expanses of shoulder-high office cubicles. Furnished with easy chairs, coffee tables, and desks, these areas serve as high-tech town squares where researchers can talk, drink coffee, and show off their latest innovations. R&D directors are trying to populate their laboratories with a new generation of researchers who thrive on the technology-transfer process: those with both the talent to work as top scientists and a passion for seeing the results of their work applied. They command big starting salaries; a survey shows the average for a newly minted Ph.D. in electrical engineering is $56,000. The leading companies track potential employees just as sports scouts do. ''My recruiting isn't much different from that of a football coach,'' says TI's Heilmeier, who devotes 10% of his time to the task. ''You can't just show up when a guy is about to graduate and say, 'Hey, come work for us!' We identify the best students at top technical universities two years before they get their degrees. We talk to their thesis advisers, visit them at school, bring them here for summer employment, and take their wives out to dinner -- because their wives decide where they're going to live. I tracked one guy for three years -- I knew his wife, what she liked and didn't like. I knew how they related to each other -- I knew everything.'' Once hired, researchers are counted as precious corporate assets, and bosses spend heavily to keep them loyal. Heilmeier says he pays his key employees what senior administrators get in other businesses -- to the consternation of bureaucrats in TI's personnel department, who try to maintain uniform pay scales across the company. ''I pay researchers what they're worth, independent of how many people they have working for them,'' he says. It is not unusual for a top scientist in U.S. industry today to earn $200,000 a year or more. At Du Pont, TI, Bell Labs, and elsewhere, sizable cash awards for important inventions are common. H-P gives its top researchers stock options. Says Du Pont's MacLachlan: ''If we lose people, we want to lose them in only three ways: retire, transfer to one of our industrial departments, or go to a university and become our friends.'' THE BEST COMPANIES supplement their R&D by exchanging ideas and findings with other companies on a scale not seen in the past. Says Penzias of Bell Labs: ''We won't invent everything ourselves -- that's guaranteed.'' One reason for last month's visit to H-P was to lay the groundwork for technology swaps. Big companies are taking advantage of a 1984 liberalization of antitrust law that allows businesses to pool their R&D. For example, IBM, Bell Labs, and the Massachusetts Institute of Technology recently formed a consortium to commercialize high-temperature superconductivity, the technology in which IBM scientists won the 1987 Nobel Prize. Corporate research giants are also linking with startup companies. For instance, specialists in ceramics and other fields at Du Pont work closely with Lanxide Corp., a small, privately held business in Newark, Delaware. Corporations have also begun to work more intensively with U.S. government labs -- a vast complex, devoted mainly to military research, that operates on a $20 billion annual budget. Lately H-P and Du Pont have begun to coordinate their efforts in high-temperature superconductivity with the Los Alamos National Laboratory. The companies will supply expertise in the application and fabrication of experimental materials, while Los Alamos will contribute its knowledge of theoretical physics as well as time on its mighty supercomputers, which it keeps on hand primarily for nuclear weapons research. Many corporations are expanding their foreign laboratories and reaching out with new types of international alliances. H-P has set up a Japanese laboratory in conjunction with the University of Tokyo. It has also opened a science center in Italy at the University of Pisa, where researchers from H-P will work side by side with their academic counterparts. AMBITIOUS LINKS with the outside world seem sure to bring U.S. corporate labs into closer contact with the marketplace. Hewlett-Packard encourages researchers to help promote the products they invent -- physicist Steven Newton, for example, took time off to write marketing brochures for an instrument he had helped invent (see box). Similarly, salesmanship is becoming a way of life at Bell Labs. Arno Penzias recently spent a day consulting with McDonald's, which uses AT&T's advanced digital network to speed communications within its Oak Brook, Illinois, headquarters complex. There could be no more vivid evidence of the way American research has changed than for the discoverer of the Big Bang to be sharing his insights with the makers of Big Macs.