What Made Bell Labs Great

HALF A CENTURY AGO AMERICAN PHYSICISTS created two of our era’s most important inventions. At Los Alamos, New Mexico, the Manhattan Project built the atom bomb. At Bell Telephone Laboratories a much smaller group of scientists made the transistor. The director of the latter project, Mervin Kelly, parlayed his achievement by building Bell Labs into the nation’s greatest industrial research center. Until its offspring, Silicon Valley, came to the fore during the 1960s, the institution Kelly built stood alone at the top.

So prominent is Bell Labs in the field of industrial research that it tends to dominate any discussion of the topic. It can be cited in support of virtually any point one cares to make: the importance of mission-oriented research, or of untrammeled scientific inquiry; the need for a strong central figure, or for loose reins of authority; the gains that can be made by huge organizations working together, or by lone researchers; the critical role of government and military funding, or of private enterprise.

In fact, the Bell Labs story demonstrates all these things. Most of all, it shows what can be accomplished with an all-out technological assault on .a broad (but not unlimited) domain that is ripe for innovation—or, more precisely, that constantly and urgently demands innovation. It also shows the advantages of working on a grand scale where small improvements can add up to large profits, and the importance of leadership from a visionary director. Unlike, say, 3M, where a relatively small research staff was encouraged to pursue any direction that seemed promising, Bell Labs stuck to a single field—telecommunications—but used its size and privileged market position to explore that field to its furthest recesses. This approach was the key to its dominance—and to why that dominance didn’t, and perhaps couldn’t, last.

ALTHOUGH BELL LABS WAS NOT OFFICIALLY established until 1925, its parent company, American Telephone and Telegraph (AT&T), showed a strong commitment to research beginning in the 1880s. Improving the transmission of voices was one area of effort, of course, but once the network expanded beyond Alexander Graham Bell and Mr. Watson, providing for connections between any pair of users turned out to be an even bigger problem. High-quality telephone service proved to be demanding, and as the telephone caught on, AT&T’s chief headache shifted from obtaining more customers to servicing the ones it had. In response, as early as 1903 the company announced a new policy: to buttress its corporate position not through patent litigation but by pursuing superiority in science and technology.

An early source for such superiority was the physicist Robert Millikan at the University of Chicago. He was one of the first Americans to win the Nobel Prize, in 1923, and he went on to help found the California Institute of Technology. Millikan did not work for AT&T, but together with his colleague and fellow Nobelist Albert Michelson, he produced a steady stream of Ph.D.’s, many of whom promptly entered the Bell System. These included Frank Jewett, the first president of Bell Labs; Harold Arnold, who laid groundwork for the first coast-to-coast telephone line; Karl Darrow, editor of the Bell System Technical Journal ; and Mervin Kelly, the biggest figure of all in the lab’s history.

Long-distance telephony was an early focus for Bell’s scientists. Just before World War I Arnold made important improvements in the vacuum tube that allowed it to be used as an amplifier. By 1915 Bell and Watson could re-enact their famous inaugural phone call of forty years earlier, this time with Bell in New York and Watson in San Francisco. (When Bell said, “Come here, I want you,” Watson replied, “It will take me five days to get there now!”) Unfortunately, when long-distance lines had to use many such tube amplifiers, voice quality deteriorated.

A staffer in the first years of Bell Labs, Harold Black, hit on the solution while riding a ferryboat to his office in New York City on an August morning in 1927. He realized that he could build a distortion-free amplifier by tapping off a small part of the amplified signal and subtracting it from the input signal. He sketched a diagram and wrote some equations on a page of The New York Times . In time those initial notes would grow into much more than an important advance in telephony. Black’s work would usher in the theoretical understanding and formal treatment of a new and major engineering field: feedback-control systems.

While Bell Labs was making its name as an engineering center par excellence, its commitment to basic research remained hit-or-miss. In 1922 Bell had begun publishing the Bell System Technical Journal , which gained a strong reputation among physicists. One early milestone came in 1927, when Clinton Davisson, a staff physicist, showed by experiment that electrons could behave as waves as well as particles. A decade later this discovery would win Davisson his own Nobel in physics, the first such prize to go to a scientist in an American industrial laboratory.

Davisson’s assigned research had stemmed from a patent dispute involving vacuum tubes. His work on electron waves reflected personal curiosity, but he had had to persuade management to allow him to pursue those studies, because no Bell Labs policy granted him freedom of research. Much the same happened a few years later, in 1932, when a staffer named Karl Jansky opened a new field of investigation in science and found himself barred from it outright.

Jansky’s official assignment grew out of long-distance radiotelephony, which transmitted messages across the Atlantic by bouncing signals off the ionosphere. The received signals were weak, demanding a sensitive antenna, and Jansky set out to discover the sources of static and raise the sensitivity by eliminating them. He found that some of the static came from interstellar space, and with this discovery he invented radio astronomy. But that topic lay beyond the domain of Bell Labs, and management barred Jansky from further research in the field.

NEVERTHELESS, BY 1930 BELL LABS HAD TECHNICAL strength that was both broad and deep in certain specific fields of physics and engineering pertinent to electronic communications. It also had great financial strength, for it operated as the research arm of the Bell System, which held a near-monopoly on telephone service. But it did not stand out among top-of-the-line industrial research centers.

At RCA the president, David Sarnoff, was supporting the work of Vladimir Zworykin, an inventor of television (who had already made key advances in the 1920s at Westinghouse). At Du Pont, which had traditionally dealt in gunpowder and munitions, the chemist Wallace Carothers was creating nylon and other plastics. General Electric, a builder of generators and turbines, had Sanford Moss, whose turbochargers would greatly increase the speed and altitude of piston-powered aircraft. Moss’s colleagues would take the lead in building America’s first jet engines. Another GE researcher, the chemist Irving Langmuir, went on to win a Nobel Prize for his researches in surface chemistry. In the 1930s these companies and others rivaled Bell Labs for first place in the industrial-technology standings.

Within two decades, though, the transistor would bring Bell Labs far greater prospects and renown. Bell’s road to the invention began quite modestly, with vacuum tubes and electromagnets. Vacuum tubes had been important to Bell since early in the century, and by the early 1930s a department with two hundred employees was working to improve them. The tubes relied on thermionic emission, in which electrons boil off a hot filament to be received by a metal plate, but the physics behind both parts of the process was little understood. In addition, every telephone receiver contained an electromagnet, so other Bell Labs physicists were interested in the magnetic properties of materials.

Another topic of interest was semiconductors. As the name implies, these differed both from metals, which readily conduct electricity, and from insulators, such as glass and rubber, which do not conduct it at all. Semiconductors ordinarily resembled insulators but could be made to conduct, after a fashion, by heating them or shining a light on them. Such materials had been known for a hundred years and had another Useful property, which Germany’s Ferdinand Braun had discovered in 1874: They could rectify, conducting electric currents in one direction only. This had made them important in radio.

During the earliest days of radio, many hobbyists built crystal receivers. The crystal was a lump of galena, or lead ore, a semiconductor; it picked up the radio signals, which could be heard through earphones. The development of vacuum tubes made crystal sets obsolete, for tubes could not only detect the signals but amplify them as well. But radios still needed rectifiers, and in 1926 two investigators at Westinghouse invented one made with an oxidized plate of copper. A Bell Labs manager, Maurice Long, conducted experiments and realized that some characteristics of this rectifier “were so similar to those of vacuum tubes that we ought to find out what made them tick.” In 1929 he hired Walter Brattain, an experimentalist, to study them. Brattain would eventually become an inventor of the transistor.

SINCE ITS EARLIEST DAYS THE BELL SYSTEM HAD CON centrated its research in the field of electrical science. Now more and more topics were being included under that rubric. Rheostats, circuits, and capacitors had been joined by thermionics, electron-metal interactions, and the fundamental properties of semiconductors and magnetic materials, all of which were at the very frontiers of physics. They demanded theoretical understanding, and the means to such understanding seemed to lie in the rapid development of quantum mechanics. Researchers in Germany took the lead in using quantum mechanics to gain fundamental understanding of the thermal, electrical, and magnetic properties of crystals. At Bell Labs this work brought the dazzling prospect that scientists might make new and basic contributions to telephony by working from first principles in this new field.

With its offices in Manhattan, Bell Labs lay close to the piers where Europe’s leading physicists arrived when visiting the United States, and several of them stopped by to give seminars. Meanwhile, the editor of the Bell System Technical Journal , Karl Darrow, tried to introduce his colleagues to the new developments by writing a series of survey articles. Brattain later said that those essays, in which Darrow tried “to explain the newer things in physics in his gorgeous language,” had first made him aware of Bell Labs.

Other Bell staffers struggled to learn the new science by studying textbooks. “We really worked,” said one of them. “We met once a week after having covered a chapter each week, and having done all the problems on it, met to correct each other’s problems. We worked good and hard.” Senior managers encouraged such study, arranging for some staff members to take courses at Columbia University, a subway ride uptown. However, mastery of quantum mechanics demanded more than courses and self-study. It called for the intense focus of graduate work leading to the Ph.D.

In addition, Bell’s knowledge-hungry staff found that physical theory still was not well enough developed to solve practical problems. For instance, Alan Wilson, a British physicist, had devised a theory of the behavior of electrons in semiconductors. But as the historian Lillian Hoddeson would note, “as late as 1938 the direction of the rectification of semiconductor-metal junctions, predicted by the best version of Wilson’s theory, was opposite to that observed.”

Into this confusing situation stepped a new director of research, Mervin Kelly, a Millikan Ph.D. Frank Jewett, yet another former Millikan student who would soon head Bell Labs, had recruited him as a staff physicist in 1918. Kelly headed the vacuum-tube department from 1928 to 1934 and took over the entire research effort in 1936.

Kelly was known for a fiery temper, triggered by anything he perceived as slackness. When provoked, he would turn dark red, though he quickly cooled down. A senior executive stated that when Kelly became director of research, the managers reporting to him were those who could take his personality and so protect those at lower levels. A Bell Labs chairman described him as “a person of limited patience, for he often regarded patience as an obstacle to action.” And an AT&T board chairman recalled that “Mervin was always and forever pushing the operating management, and the heads of AT&T as well, to get on with new things. His aggressiveness got him in a lot of hot arguments, but I always sat back and said, ‘Give it to them, Mervin, that’s what we need.’ Every place needs a fireball or sparkplug, and he was it.”

He knew how to listen, too, and would admit when he was mistaken. His memory was phenomenal; one staffer wrote that “after someone had shown and explained his work, Kelly would remember everything a year later.” His assistant said that he “had very good judgment in deciding which things people were working on were boondoggles and which were fundamental.” And though he was an organization man, he prized the individual, asserting that “it is in the mind of a single person that creative ideas and concepts are born.”

He had his own hopes for semiconductors, for he thought they could serve as solid-state switches. The telephone exchanges of the day relied on electromechanical relays, which were costly to maintain. By contrast, semiconductor switches would have no moving parts. Kelly felt strongly that Bell’s scientists needed basic understanding and were relying on rules of thumb in making their semiconductors. “Processing of selenium rectifiers was largely an art—cookbookery,” he wrote. For example, “we used only the copper from a particular mine for fabricating” certain semiconductor devices. “It made the best. We did not know why.” Kelly set out to hire young Ph.D.’s who could work in solid-state physics.

At the time, amid the Depression, physics held a marginal place in American society. The historian William Manchester writes that as late as 1939, on the eve of World War II, “the very word ‘physicist’ was uncommon; many Americans couldn’t even pronounce it. Universities paid men with Ph.D.’s in science $1500 to $1800 a year. They accepted because they had little choice. Industry didn’t want them. In one year, 1937, there were only four research openings for them in the whole country.”

But solid-state physics was taking hold at major universities, with Eugene Wigner at Princeton, John Slater at MIT, and John Van Vleck at Harvard leading the way. Kelly proceeded to recruit students from these institutions and elsewhere. From Cal Tech he hired John Pierce, who would go on to invent the communications satellite, and Dean Wooldridge, who would become a founder of the firm now known as TRW. MIT yielded William Shockley, a student of plater’s, who would participate in inventing the transistor. Shockley found Bell Labs attractive because he knew of Clinton Davisson’s work, and when Davisson won his Nobel Prize in 1937, Bell looked more attractive still. Kelly went on to hire MIT’s James Fisk, another student of Slater’s, who rose in time to become president of Bell Labs, and Charles Townes of Cal Tech, who would win a Nobel for developing the theory of the laser.

As the 1930s progressed, the pace of discovery in semiconductors picked up worldwide. One line of research emerged in Germany at G‶ttingen University, where Robert Pohl had predicted that radio vacuum tubes would be replaced by small crystals in which the motion of electrons could be controlled. In 1938 he went on to create such a crystal, of potassium bromide. It offered no route to a practical device, for it relied on thermal effects that were hard to control. But it showed that a semiconductor could serve as an amplifier in an electronic circuit. It was an early demonstration of the principle of solid-state amplification that would reach fruition in the transistor.

At Bell Labs, Shockley and Brattain were alert to such possibilities. Their early efforts produced no breakthroughs, but they whetted the scientists’ appetites, and elsewhere semiconductor studies began to achieve practical success. The first major benefit was the thermistor, a device whose electrical resistance diminished with temperature—the opposite behavior from that of conventional resistors. Using both types, it was possible to compensate for the effects of heat and cold on long-distance phone lines.

MICROWAVE RADIO OFFERED OTHER POSSIBLE APPLI cations. Microwave technology would turn out to be essential for television and radar, and within the Bell System it promised communications channels that could each carry hundreds of separate long-distance phone calls. But microwave radio circuits needed rectifiers, and vacuum-tube rectifiers could not handle the short wavelengths. George Southworth, a Bell Labs radio engineer, thought he might get somewhere by returning to the oldfashioned crystal radio. He found what he needed in the late 1930s in a secondhand shop on Cortlandt Street in lower Manhattan and soon showed that such crystals were far better than vacuum tubes for dealing with microwaves. In effect his crystals were semiconductor rectifiers.

Southworth’s discovery sparked a substantial effort to find the best semiconductor materials and use them in wartime radar units. A group at Purdue University made a key contribution by showing that silicon and germanium offered noteworthy promise, and by preparing germanium in pure form. At Bell Labs the chemist Russell OhI tested more than a hundred crystalline materials and decided that silicon was best. He also decided that he could get even better performance by working with purer samples of silicon.

Producing such samples proved demanding, and Ohl arranged for two Bell Labs metallurgists to help him. As the levels of impurities diminished, the researchers found dramatic improvements in the samples’ properties, including the photoelectric effect, wherein one could produce a flow of current by shining a light on a semiconductor. Brattain recalled that during a demonstration in 1940 “we were flabbergasted. The effect was more than ten times greater than we had been getting.” Subsequent work brought the discovery of a crucial principle: By adding minuscule quantities of selected elements to molten silicon or germanium, one could create new semiconductor alloys with predictable, reproducible electronic characteristics.

After the war the way lay open to a concerted attack on the semiconductor problem. The work done so far had involved no commitment to unfocused research or to curiosity pursued for its own sake; still less had it involved the creative acts of geniuses. Indeed, as Shockley and Brattain had learned in their attempts to build a semiconductor amplifier, all too often they still did not know what they were doing. But the research effort drew strongly on the sheer size of Bell Labs, which was large enough to pursue a broad range of semiconductor studies. This size also meant that a variety of talents were available. When OhI needed help from metallurgists, for instance, they were there.

Mervin Kelly’s leadership was also important. Before the war Kelly had realized that he needed the specialized knowledge of experts in quantum mechanics and solidstate physics if his researchers were to turn laboratory curiosities into practical devices. He had hired those experts and left them to their own devices. Then wartime work on rectifiers brought a substantial knowledge base on semiconductor materials. In the words of John Pierce, “Without these advances in materials, you wouldn’t have stood a ghost of a chance of making a transistor.”

Kelly was well aware that a number of other labs, in the United States and Britain, had been working on semiconductors for radar. He thus expected fierce postwar competition in the broad field of semiconductors, and in July 1945 he set out to meet it by reorganizing Bell Labs’ physical-research department. It had continued to emphasize topics pertinent to vacuum tubes, but Kelly now made solid-state physics the principal topic of investigation.

AS EARLY AS 1938 KELLY HAD GONE EVEN FUR ther by setting up a small group that included Shockley and Wooldridge and giving them unprecedented freedom to chose their own problems within the semiconductor field. Kelly had no intention of setting Bell Labs’ entire staff to work on what- ever flight of fancy a physicist might dream up, but he recognized the importance of letting a few key men with insight choose their own agenda. When they came up with something promising, Bell Labs would give them the resources they needed.

In 1940 Kelly had written a report to his management, stating that such free-form research would not “bring immediate results applicable to our business. However, its method of approach is so basic and may well be of such far reaching importance that we should have such studies in progress.” After the war, in Lillian Hoddeson’s words, “the number of seminars, journal clubs, and study groups increased, and efforts were also made to provide further education for junior staff. These attitudes and circumstances, together with easy access to a larger staff of scientific and technical specialists in many fields, made the research environment for some Bell scientists superior to that offered at the best universities.” These represented practical steps aimed at strengthening the lab’s ties to the scientific community.

Drawing anew on Bell Labs’ wide range of knowledge, Kelly carried through his 1945 reorganization by setting up a number of small research groups that would operate as interdisciplinary teams. Shockley, recently returned to Bell Labs from his wartime research, headed one such group. He was a theorist in solid-state physics, but he knew he needed help in this area, and he persuaded Kelly to hire a fellow theorist, John Bardeen, who had worked with Wigner at Princeton and Van Vleck at Harvard. Shockley’s group also included Walter Brattain and Gerald Pearson, experimental physicists who had worked on semiconductors for more than a decade. In addition, the group included a physical chemist, an expert on electronic circuitry, and two technical assistants. Few other labs could have matched this breadth of experience, and few universities had the flexibility to assemble such a varied group so easily or were geared to develop inventions.

Shockley still wanted to invent a semiconductor amplifier, and he tried to do so by relying on a phenomenon called the field effect. He would mount a sliver of semiconductor on a metal plate, and by applying a weak voltage to the plate, he hoped to add electric charges to the sliver, changing its conductivity and allowing it to modulate, or control, a much stronger flow of current passing through it. Shockley was quite surprised when his approach failed to work, but at this point the theorist Bardeen stepped in to explain why: Rather than accumulating within the interior of the semiconductor, the electric charges were being trapped at its surface.

Bardeen’s theory opened a different line of work. Instead of fighting this surface accumulation of charge,) the team built an amplifier that took advantage of the effect. This effort succeeded, and in late 1947 the researchers demonstrated the first experimental transistor. It was the “pointcontact” type and bore some similarity to the old leadore crystals in early radio sets, for it depended on having thin and closely spaced wires making contact with appropriate spots on the surface. It nevertheless was a true semiconductor amplifier, and it opened the way to the solid-state world that would soon emerge. Shockley made another important contribution in 1949 with the junction transistor. It proved to be more robust and less sensitive than the point-contact type and emerged as a standard.

The earliest transistors were expensive and tricky to work with. As one manager put it, “The transistor in 1949 didn’t seem like anything very revolutionary to me. It just seemed like another one of those crummy jobs that required one hell of a lot of overtime and a lot of guff from my wife.” But even then its promise was immense. The vacuum tube was bulky, fragile, limited in life, and wasteful of power; the transistor offered minute size, ruggedness, long life, and immediate action with no need to warm up. Best of all, it promised very low power consumption, and hence little production of heat. In turn this meant that computers could come into their own, compact in size and reliable in operation.

The first electronic computer, ENIAC, had contained some 18,000 vacuum tubes and consumed 140 kilowatts of power. Although it pressed the limit of what-could be built with such tubes, it spurred far-reaching thoughts of future, more powerful computers. Transistors, overcoming the limits of those tubes, would make it possible to build them. As early as 1953 Fortune magazine predicted that transistors would “lift information-handling and computing machines—the nub of the second industrial revolution now upon us—to any imaginable degree of complexity.”

HOW HAD KELLY LAUNCHED THIS SECOND INDUSTRIAL revolution? The invention of the transistor had drawn on fundamental physics to a degree unprecedented in the history of industry, and Kelly had developed his research department in ways that were equally unprecedented. At a time when no one else wanted to hire physicists, he had gone out of his way to seek the best and give them an environment where they could grow. This gave Bell Labs a reputation as a place where physics was welcome, a reputation that would make it easier for Kelly to continue to hire excellent researchers.

In addition, he had supported these people by promoting freedom of research within an academic atmosphere and by placing the extensive resources of Bell Labs at their disposal. In particular, his research teams broke down barriers between disciplines, as physical theorists worked with experimentalists, with chemists, metallurgists, and circuit designers close at hand. As always, the research was aimed at ultimately creating products that would advance the field of electronic communications. But the quality of Kelly’s best specialists, the depth of understanding that they gained through his policy of academic freedom, and the breadth and power of his research groups—all these things marked Bell Labs as a place unto itself.

Indeed, Kelly had gone far toward creating a new type of research center that could combine the best features of the university and the traditional industrial research lab. Universities offered academic freedom galore, but if scientists wanted to build something and market it as a product, they were out of luck. They might conduct experiments, but the necessary supporting staff and facilities wouldn’t be available.

Industrial labs, by contrast, often were too product-oriented for their own good. The best of them were true research centers, but their managers faced a continuing temptation to turn their distinguished specialists into product-development directors. If scientists pursued research at the frontiers and came up with something promising, senior management often would tell them to turn away from continuing that research and focus instead on developing the invention into a salable product.

Kelly would have none of this. As he told Britain’s Royal Society in 1950, the best industrial researchers “must be given freedoms that are equivalent to those of the research man in the university. It is most important for the scientists to confine their efforts to the area of research. If they extend the area of their effort even to that of fundamental development [the area of work that immediately follows research], they tend to lose contact with the forefront of their field of scientific interest. In time, a considerable fraction will lose their productivity in research.”

And Kelly practiced what he preached. Following the invention of the point-contact transistor, for example, Kelly did not tell Shockley to redirect his work toward further developing and refining the product. Instead he gave that effort to another manager and left Shockley free to seek newer frontiers. Shockley responded with the junction transistor.

Still, many staffers found these policies less absolute than they seemed, even after Kelly became president of Bell Labs in 1951. As one scientist put it, “The humble Ph.D.’s … felt like cogs in a very large machine.” In 1954 Business Week wrote that “partly, the freedom is illusory. The lab has firm plans, and knows precisely what it wants. Over the years, personnel has been meticulously selected, and precisely trained. Men chosen to fit a mold will fall into the desired pattern without any pressure from the mold itself.”

Even so, for a modest number of world-class researchers, Bell Labs could indeed be all it promised. It could improve on a university by freeing scientists from attending committee meetings, teaching undergraduates, and begging for grants. For these reasons Claude Shannon came over from MIT to create a new science—information theory. Two other Bell physicists, Arthur Schawlow and Charles Townes, elucidated the basic theory of the laser. At about the same time, other researchers assembled the first working solar power cell.

In 1956 Shockley, Bardeen, and Brattain received their Nobel Prizes. By then the reputation of Bell Labs, already beyond peer in the engineering world, stood alongside that of Cal Tech or MIT in the world of science. The nation’s leading physics journal, the Physical Review , published sixty-six papers from Bell Labs during 1956 and 1957. This total was exceeded only by Berkeley, MIT, Chicago, and the national laboratories at Brookhaven and Los Alamos. One head of research said that it was easy to assess a staff scientist’s value: “We just ask, is the field his; does he own it; did he write the book? Are other people thinking with his thoughts?”

YET IN THAT SAME YEAR OF 1956, BELL LABS’ LEADER ship began to slip irretrievably. Facing antitrust action from the Justice Department, the Bell System entered into a consent decree by which it surrendered its existing transistor patent rights. In addition, Bell Labs agreed to work only in the field of telecommunications. The consent decree barred it from the field of computer equipment.

Bell Labs had gotten off to a good start in computers while striving to let customers dial long-distance calls directly, without operator assistance. This required equipment that could receive a dialed number, including the area code, store it in memory, then draw on the information in the number’s digits to determine the destination. The system had to set up the connection, picking out the shortest route, testing it to see if it was busy, and then choosing an alternative route if necessary. It had to make sure it had the correct number, operate all necessary switches, and then disconnect from the circuit to take care of other customers. Meanwhile, other system elements had to time the call and keep track of the caller and destination to allow for automated billing.

The first such system went into operation in 1951. It amounted to a special-purpose computer with hard-wired logic, operating arrays of electromechanical switches. During subsequent decades Bell Labs maintained an active involvement in computer technology, as it built increasingly capable direct-dialing systems and switching equipment. But under the consent decree it could not compete in the general computer market. Instead it maintained its strong involvement with applied physics.

This corporate strategy was motivated by more than the consent decree, for it played to Bell’s strength. Bell’s pathbreaking advances had required little capital investment but had yielded a flood of patents. Since AT&T had a virtual monopoly in telephone service, any improvements in that field were sure to pay off; there was no risk of being beaten to the punch or losing staff to a competitor. By contrast, a serious effort in computers would have brought Bell face-to-face with IBM. Such a commitment, even if legally permissible, would have required vast investment, with excellent prospects for major losses. For IBM owned the computer field.

IBM’s unquestioned strength, which dated to the mid1950s, did not result from superior electronics technology. Rather, it grew out of a long-standing and strong commitment to customer support. This had won IBM a large measure of loyalty even in the precomputer era, when in addition to typewriters it had built such machines as calculators and card-reading printers. This policy carried over to the computer era, allowing IBM to defeat such heavyhitting competitors as RCA and General Electric. The same fate would likely have befallen AT&T if it had chosen to challenge Big Blue.

By contrast, within its chosen field of telecommunications, Bell Labs found ample play for its talents as it worked to introduce telephone service that would replace the traditional copper wires with optical fibers. Lasers emerged as an early and important interest. During 1958 Arthur Schawlow and the consultant Charles Townes published their fundamental theory of laser action. Two years later Ali Javan invented the first continuously operating laser. And when available glasses proved inadequate for optical fibers, Bell Labs scientists came up with their own. In a separate development John Pierce, who had joined Bell Labs in 1936, invented the communications satellite. The first commercial version, Telstar , reached orbit in 1962.

Meanwhile, Bell Labs faced a separate challenge. In a switch from the hardscrabble Depression days, a seller’s market for science Ph.D.’s had blossomed after World War II, as both military and civilian demand for technological expertise exploded. The start-up firms of a burgeoning Silicon Valley offered opportunities that Bell at its best could not match. In the 1920s and 1930s new graduates had been grateful to find any job related to their field. Now they had dollar signs in their eyes, and while outstanding scientists at Bell might enjoy academic freedom, they lacked the freedom to grow rich.

The Bell System belonged to its stockholders, and though AT&T was a prime investment, no entrepreneur could hope to own a significant portion of the company. In addition, Bell scientists had to sign all patent rights to their inventions over to the company; in Silicon Valley, on the other hand, scientists launching a start-up could enjoy both corporate ownership and patent rights. They could legitimately hope to become millionaires, and many did just that. As time went by, this handicap put Bell Labs at an increasing disadvantage. Its reputation and resources would still attract good people, but they would be those who lacked the moxie to launch a start-up or whose interests did not involve the development of computers.

The new start-ups showed their strength as early as 1958, with the invention of integrated circuits by Robert Noyce and Jean Hoerni at Fairchild Semiconductor and by Jack Kilby at Texas Instruments. Both firms had drawn from the strength of Bell Labs: Fairchild had been founded by colleagues of William Shockley, who left Bell in 1955 to return to his hometown of Palo Alto, California; Texas Instruments, working in the realm of commercial transistors, had relied on a provision in the 1956 consent decree that gave it ready access to Bell’s patents.

After that, while Bell Labs triumphed with its lasers and communications satellites, milestones in electronics came increasingly from the PaIo Alto area. The first microprocessor appeared in 1971, from the new firm of Intel. Six years later Steven Jobs and Steve Wozniak founded Apple Computer. In 1981, when IBM needed microprocessors for a new line of personal computers that would become an industry standard, it bought them from Intel.

Within its legally permitted fields, Bell Labs continued to stand out. These included more than solid-state physics and telecommunications: they extended to computer software. Bell Labs introduced the Unix operating system and the computer language C, along with early programs for word processing. In accordance with the consent decree, Bell gave them away free to universities, which made Unix and C highly popular among the leather-elbow crowd. Even here, though, Bell lost out to such start-ups as Microsoft, which gained dominance in software to rival that of Intel in microprocessors.

Then in 1984 a new legal decision broke the nationwide Bell System into seven regional operating companies. This decision repealed major portions of the 1956 consent decree, allowing Bell to enter the world of computer equipment. AT&T responded by merging with NCR, which had been active in this area, but the marriage did not last; the firms split up again in 1995. Bell Labs, as well, fissioned in 1984. Some 3,000 of its 22,000 employees formed the nucleus of a new firm, Bell Communications Research (Bellcore), which would concentrate on meeting the Baby Bells’ needs.

THESE CHANGES RAISED A SERIOUS QUESTION: Could Bell Labs maintain its commitment to excellence? The breakup of the Bell System kept Bell Labs as an arm of AT&T, protecting its funding and ensuring that its budget and staff could continue to grow. Still, the lab’s commitment to basic research had been built on its ability to take the long view, in the expectation that the Bell System would continue indefinitely as a regulated monopoly. Accordingly, the lab’s researchers had felt free to look beyond the problems of the moment, pursuing studies that might turn into useful technology decades in the future, if ever.

Their researches had had to be related to telecommunications, of course, but Bell Labs could afford to construe this mandate broadly. It could develop equipment for the communications systems of the future, because it was a safe bet that AT&T would still be running those systems. Kelly’s support of solid-state physics, years before the transistor lay in sight, had reflected this long view. In similar fashion, work on lasers and optical systems, which had begun in the 1950s, took twenty years to culminate in major projects that spanned the nation with fiber-optic communications during the 1980s.

But the 1984 breakup thrust AT&T into a new and highly competitive environment, in which it would have to struggle for market share like any other large corporation. Amid these new circumstances its ability to hold the long view would not be guaranteed. The vice president for research, Arno Penzias, had won a Nobel in physics, and he kept up the old ways as long as he could. In 1990, though, he carried through a major reorganization to make Bell Labs more responsive to near-term business needs. The era of university-style research was over. A second reorganization, in 1995, continued the fragmentation process and further distanced Bell Labs from what it had been at mid-century.

Like much of American business, AT&T seems to be taking the small, nimble firms of Silicon Valley as a model. In so doing, it has moved away from its former broad research policies and adopted, for better or worse, the tight focus on product development that characterizes such companies as much as their record of innovation.

Some leading scientists voted with their feet. The University of Illinois, a leader in solid-state physics, found itself flooded with applications from Bell Labs. Robert Dynes, a top superconductivity expert, left for the University of California at San Diego. He remarked that when he was at Bell, “all you had to do was really good research and be a leader in your field. Now you can still do good basic research, but the really important yardstick is its relevance for the company’s business.”

Charles Shank, who left after the elimination of an electronics lab he had headed, had his own view: “Fundamental new advances come over time, and if you’re going to invent something like the transistor or the laser it requires an organization with size, not a start-up company. The single most important thing to a thriving basic research lab is stability in terms of long-term commitment of resources. That’s what creates a scientific culture, and it was the key to the success of Bell Labs.

“You can’t build up a scientific culture quickly, but you can sure tear it apart in a hurry. That’s what we’re seeing here. And the tragedy of this whole story is that American society hasn’t realized what it’s lost.”