by George Dyson
Arthur Burks, Goldstine’s colleague from the ENIAC project, was hired on March 8, 1946 (at a salary of $4,800), and electrical engineer James Pomerene was hired (at a salary of $4,500) on March 9. Twenty-six years old and recently married, Pomerene showed up for work on April 1, and was soon joined by Hazeltine colleague (and Moore School alumnus) Willis Ware, who accepted a position on May 13 and began work on June 1. Pomerene and Ware took the train down to Princeton to meet Bigelow. “We drove back to New York City with Julian in his little old green Austin,” Ware recalls. “By the time Pom and I got back to New York City—we were still at Hazeltine—we were absolutely enthralled. He just bubbles with ideas all the time.”27 Pomerene and Ware were able to trade their New York apartments with two Princeton residents who were working for the United Nations in Manhattan, exchanging a long commute by train for a short commute by bicycle down Nassau Street and Olden Lane.
Pomerene and Ware had both worked on pulse-coded IFF (Identification Friend or Foe) radar systems during the war. As soon as radar had made it possible to hit targets at night or beyond visible range, otherwise adversarial air forces agreed on a system of coded signals identifying their aircraft as friend or foe. In contrast to the work of wartime cryptographers, whose job was to design codes that were as difficult to understand as possible, the goal of IFF was to develop codes that were as difficult to misunderstand as possible. Pomerene and Ware (and their counterparts Frederic C. Williams and Tom Kilburn on the British side), who had developed circuitry to communicate coded pulses, at high speeds across noisy channels between aircraft, now faced the same problem in building electronic digital computers: how to transmit coded pulses, thousands of times per second, from one machine cycle to the next. We owe the existence of high-speed digital computers to pilots who preferred to be shot down intentionally by their enemies rather than accidentally by their friends.
A small team began to coalesce. Richard W. Melville, a seaman first class technical specialist in radar, “came in with a sailor’s hat on, and asked for a job,” according to Bigelow, “and I liked him and he seemed to be on the ball.”28 Melville turned out to be “a wizard,” in Pomerene’s assessment, at improvising the requisite laboratory facilities, while supervising the mechanical engineering of the machine. He kept everything running smoothly in tight quarters, worked miracles at finding war surplus materials and parts, and hired mechanically inclined high school students once designs were finalized and forty-stage copies of the prototype shift registers and accumulators had to be produced. His wife, Claire, taking over a vacant apartment, opened a nursery school for Institute children too young to attend the Princeton public schools.
William S. Robinson, mechanic, was hired on March 21, 1946, and was joined in the machine shop—once there was one—by Winfield T. Lacey and Frank E. Fell. Ralph Slutz, a graduate student in physics at Princeton, accepted an offer on April 5 to start full-time work on July 1. “I went and knocked on John von Neumann’s door,” Slutz remembers, “and said ‘I hear you’re going to build a computer, how about a chance to work on it?’ He said yes.” Slutz had met von Neumann in the course of wartime work on blast waves, and heard about the prospects for doing computation with vacuum tubes. “I remember sitting in my classes,” he recalls, “sketching out adders rather than paying any attention to quantum mechanics.”29
Robert F. Shaw, one of the ENIAC veterans still at the Moore School, accepted a position as a member of the engineering staff on May 13. John (Jack) Davis, also at the Moore School (and former neighbor and high school classmate of Willis Ware), accepted an offer on April 13 to report to work on June 1. “I used to sit on Jack Davis’ bed and listen to the short wave radio on receivers that we’d built on our mothers’ pie pans,” remembers Ware.30 Ames Bliss, son of ballistician Gilbert Bliss, accepted a position as contracts administrator on May 14, at a salary of $4,000 per year. Akrevoe Kondopria of Philadelphia, Goldstine’s secretary at the Moore School, transferred to the IAS and reported for work on June 3, 1946.
“I was sixteen and came from a Greek immigrant family. My father was from a poor Ionian island and could barely read and write. It was made quite clear to me that I couldn’t go to college, much as I wanted to,” she remembers. Her guidance counselor advised forgoing college and applying for a job as a secretary at the Moore School of the University of Pennsylvania. “There I met Captain Goldstine, with his two gold bars, slim and elegant in his uniform, and Mrs. Goldstine, Adele, who dressed casually and smoked a lot. For some reason they hired me. I, who had never taken algebra, was thrown into a world completely foreign to me, and it changed my life.”
“The Goldstines were wonderfully encouraging, and as the time grew nearer to move to Princeton, they asked me to come along with them,” she explains. She commuted by train from Philadelphia until offered a room in mathematician Salomon Bochner’s house, at the end of Springdale Road. Using a manual typewriter at first, and the Institute’s new Varityper later, she helped produce the progress reports that were being generated even before the construction of the machine got off the ground. “I remember how tedious it was, because you had to change the type font from text to the mathematical symbol disc, and of course, you had to be very, very accurate,” she recalls. She had no inkling, at first, of the hydrogen bomb calculations that were driving the project at IAS. “Nicholas Metropolis was always sending letters to P.O. Box 1663, Santa Fe, New Mexico, and was getting letters from the same address,” she remembers. “My guess was that he must have had a girlfriend there.”
Akrevoe remained in Princeton until August 1949, when her mother, who “thought I was getting beyond my station,” insisted she move back to Philadelphia, presenting the ultimatum as “It’s time to come back; you’ve gotten too many big ideas.” It was hard to leave. “The Goldstines and the young engineers treated me like a kid sister,” she remembers, “and probably taught me more than I would have learned in college.” The presence of a bright, red-haired seventeen-year-old was just what the computing group needed. “Some of these guys were so helpless,” she adds. “They didn’t have very many social skills.”31
Between the stuffiness of Princeton and the rarefied atmosphere at the Institute, it was not easy for the engineers to fit in. Goldstine, a mathematician underneath his army uniform, and Burks, a logician finishing up his wartime service as an electronic engineer, had been granted sanctuary in the annex to Gödel’s office on the second floor, from where they assimilated easily into the culture of Fuld Hall. When engineers began showing up, the welcome cooled. The ivory tower was full. “I have thought very carefully over the problem of disposing of these fifteen workers who are to arrive the middle of June,” wrote Aydelotte to von Neumann. “The only really useable space in our basement is that adjoining the men’s lavatory, to which you are most heartily welcome.”32
“There was no space for us, and so for the first five or six months, we were crowded into the boiler room with a few work benches we set out,” Bigelow explains. “There was not even an office for me to go to and hide, and think about circuit logic, without having people walking over my desk and crawling all over me.” All purchases of building materials, down to a single two-by-four, had to be reviewed by the Civilian Production Board. “Finished lumber was rationed because of housing shortages,” adds Bigelow, “so we purchased rough-sawn oak from a local purveyor of fireplace wood.”33
The engineers were shunned by the scholars upstairs. “The prevailing attitude among the humanists toward the notion of a laboratory at the Institute was one of undisguised horror,” remembers Bigelow. “The attitude of the mathematicians ranged from about the same extreme to some instances of mild interest; however, the extremists among them tempered their objections in accord with the universal respect and esteem enjoyed by von Neumann.”34
“We were doing things with our hands and building dirty old equipment. That wasn’t the Institute,” remembers Ware. “The coming of six engineers with their assortment of oscilloscopes, soldering
irons, and shop machinery was something of a shock.”35 When the project first started, “we were given temporary space in the second basement, surrounding the boilers,” he adds. “It wasn’t bad since it was summer and they were turned off.” The basement storage room was bare. “Our first job was to build work tables for us to work on,” remembers Slutz. “We asked von Neumann if he would pay for the paint if we painted the walls a more reasonable color than they were when we moved in. This he did.” The engineers also had to install their own wiring, and, reminiscent of Julian Bigelow’s installation of a single circuit to his bedroom, the first engineering expenditure by the Electronic Computer Project was four dollars, recorded by Bernetta Miller in April 1946, for “Electrical work.”36
“Our work benches surrounded the boilers and miscellaneous shop and laboratory equipment was stuffed in any available corner,” explains Ware. “With the coming of autumn, the situation warmed up quite considerably; so much so in fact that the group enjoyed an improvement in social status and moved to some unoccupied storerooms in the first basement.”37 As they moved up from the boiler room level to the first basement directly below the ground floor of Fuld Hall, there were vocal protests on all fronts. To the humanists, the computer people were mathematicians, and to the mathematicians, they were engineers. “Even the curiosity so natural to all scientific people was overruled by the passionate distance toward anything that might conceivably deviate from pure and theoretical thinking,” Klári von Neumann explained. Bigelow describes the situation as one of people “who had to think about what they were trying to do” objecting to people “who seemed to know what they were trying to do.”38
The mathematicians and the humanists occupied opposite wings of Fuld Hall. Their defenses were entrenched. “I have learned with some dismay that a group of electronic experts has moved into half of the basement of our wing at the Institute,” classicist Benjamin Merritt complained to Aydelotte in 1946.39 The engineers and the scholars were forced to coexist. “There would be social gatherings from time to time,” remembers Ware, “and they’d say ‘Well, I’m in mathematics’ or ‘I’m in physics,’ or ‘I’m in … what are you in?’ And then when one answered, it became clear that you were a social outcast. We were sort of fifth class citizens around there.”40
Salary disparities did not help. The engineers received $5,000 or $6,000 a year, barely enough to keep them from leaving for betterpaying jobs in industry, but more than what was paid to visiting scholars with advanced degrees. “Many of these people, who were good electro-technologists, had, for example, only a bachelor’s degree,” explains Bigelow, “whereas the Institute itself had visiting members with Ph.Ds from four or five important universities in the world, who came over here for fellowship stipends which amounted to $2,500 or $3,000. So there were really substantive kinds of jealousy arising from this.”41 The budget for the Computer Project—funded entirely by the government—soon grew larger than that of any of the existing schools.
The computer group needed to move out of Fuld Hall, and soon. With the postwar shortage of building materials, and the resistance of both the Institute’s old guard and the neighboring residents to anything resembling a laboratory, this was not easy to do. “At that time, building materials were under rationing,” explains Bigelow. “You could not go out and build yourself a house or a garage or something. If you wanted a few feet of wood, you had to have a certificate for it, much less things like workbenches and hardware and tools.”42 It turned out to be more difficult to obtain things under the Civilian Production Administration than it had been under the War Production Board. A lot of bargaining went on behind the scenes. “I just finished talking to Sam Feltman (at the Ordnance Department) regarding our contract,” Goldstine explained to von Neumann, when money was running low. “He says he received his allotment of money just this morning and hopes for War Dept. approval within 2 or 3 days at which time he will authorize Phila Ord. Dist. to give me the dough, Gott sei Dank!! In turn for this he has two favors: First, he wants a copy of the Theory of Games, which I’ll try and send him. Second, he wants help in placing his son in medical school.”43
The other problem was where to put the building. “There is a distinct feeling in the faculty that the computer building should not be on the same plot of land with Fuld Hall,” Aydelotte wrote to Herbert Maass, suggesting a location on the other side of Olden Lane, near the old barn. “Most of the Institute buildings are placed close together, connected with a path or archways,” noted Klári von Neumann. “This one was put way across a big empty field edged with tall bushes, beyond it a road and then the building; one could easily pretend that it did not belong with the rest at all.”44
Arthur Burks remembers helping Herman Goldstine and Oswald Veblen choose the site. “And we walked through the woods, but it was clear that Veblen didn’t want any trees to be cut down for the building. In the end, he picked a site which was low down, and not too far away.… He wanted the building to be one story only, so that this would not be a conspicuous building.” The site was too swampy to build on as is. “The site we have selected needs a little fill, and I notice that you people are excavating immense quantities of earth in preparation for building the stacks of the new Princeton library,” Aydelotte wrote to the university. “As far as I can see your trucks seem to be hauling the earth a considerable distance away, and I am wondering if we could buy a few truckloads of this earth.”45
The Electronic Computer Project’s contract with its government sponsors allowed $23,000 for the construction of a “temporary structure” to house the computer, whereas the bids for a building whose appearance was acceptable to the Institute community came in at $70,000. After much negotiation, it was decided that a flat-roofed, cement-block building would be constructed for $51,000, with the Institute paying an additional $9,000 for a cosmetic brick veneer, and reserving the right to add a gabled roof at a later time. “The exterior walls of the building are being veneered with brick in order to insure a weather-tight job, since we were advised that the present grade of stucco was of questionable weatherproofing quality,” Aydelotte wrote to Colonel Powell, chief of army ordnance, who questioned the cost of the brick veneer. “However, in order that there be no possible criticism of us in this connection, we had always planned to pay the $9,000 for this adjunct to the building out of Institute funds.”46
The new building would not be ready to move into until Christmas of 1946. In the interim, the engineering team prepared for the construction of the machine. A small machine shop (equipped with a lathe, drill press, and planer) was established in the boiler room, and the engineers began accumulating electronic components, instruments, and tools. “We obtained surplus components and built our own power supplies,” remembers Bigelow. “We really worked from the ground up.” Electronic components were still restricted for civilian use. “Whenever we wanted anything, we tried to get the Army Materiel Command to find it for us, and at that time there were itinerant salesmen who would buy war surplus and then go around and try to peddle it, and so we got a lot of stuff that way,” adds Willis Ware. “The Princeton machine was built through war surplus. We used whatever the army could find for us, so in a subtle sort of way that pushed the design.”47
The engineers used their personal connections and personal equipment to help build the computer, and used computer project resources for personal projects in their spare time. Everything from Einstein’s hi-fi to Herman Goldstine’s TV antenna was made in the ECP shop. “Princeton has the unique characteristic that it’s almost exactly halfway between Philadelphia and New York, so RCA designed an antenna that, with the flip of a switch, electronically pointed to New York or electronically pointed to Philadelphia,” says Willis Ware. “So we all, in our shop, with Melville’s aid, tooled up and made those antennas for a lot of people.”48 Jack Rosenberg took things one step further and made his own recordings. “He used to listen on WQXR every Saturday to Toscanini playing in New York and he would record it with tremendou
s high fidelity,” explains Morris Rubinoff. “Now tremendous high fidelity considering the antenna he had meant that you heard every bit of static, crackle and noise that came through, and you were proud that you could hear it all, it certainly was hi-fi—from 15 cycles to 20,000 or 30,000 or something.”49 With Aydelotte’s replacement by Oppenheimer, the freewheeling atmosphere began to change. “It has proven convenient, especially for those engaged in work on the computer, to avail themselves of the discount rates of the computer contract in making essentially personal purchases of radio equipment and parts,” Oppenheimer wrote to von Neumann in 1949. “However, it appears that in a number of cases we have carried the items so charged on our books for a quite improper length of time.”50
The design of the computer was determined partly by the commandments that Burks, Goldstine, and von Neumann dispensed from above (“They were our bible,” says Rosenberg)51 and partly by what the availability of war surplus components dictated from below. “We bought quantities of surplus electrical and electronic components, electron tubes, etc., rather indiscriminately,” says Bigelow. “We read through Preliminary Discussion… often and discussed the engineering task it presented with each other and with Johnny and Herman, who were already trying out exploratory coding procedures on paper.”52
From the beginning, there was tension between Goldstine, the associate director of the project, and Bigelow, the chief engineer. On almost every important question, from circuit design (with Bigelow clinging to “a weird and inchoate idea as how to build an adder,” as Goldstine complained to von Neumann) to the disposition of patent rights, the two disagreed. The chain of command was often in dispute. Only von Neumann was strong enough to intercede. “He kept Herman and I from fighting by some marvelous technique,” says Bigelow. “We got along like oil and water, or cat and dog; and von Neumann would keep this here, and this there, and smooth things over.”53