Author’s note: I began to write a history of computing at Princeton a decade ago while working in the University’s Office of Information Technology. That effort caught the attention of Michael Mahoney, professor of the history of science. We were just beginning to transform my lengthy draft into something special when, in 2008, Michael met an untimely death. I wish to dedicate this short essay in his memory.
This is a corrected version of a Web Exclusive article published with the April 4, 2012, issue. The correction appears at the end of the story.
Princeton’s computing story begins not with Alan Turing, Alonzo Church, John von Neumann, or Albert Einstein, but with Oswald Veblen (1880-1960; Ph.D., Chicago 1903), who came to the campus at the request of University president Woodrow Wilson 1879. A specialist in topology and in projective and differential geometries, Veblen taught mathematics at Princeton from 1905 to 1932. In 1926, he became the Henry B. Fine Professor of Mathematics.
His uncle Thorstein, who wrote The Theory of the Leisure Class (1899), is better known, but Oswald arguably had a more lasting impact. During World War II, Einstein urged Roosevelt to build the bomb. Years earlier, Princeton’s Veblen had pressed Roosevelt to help him bring mathematics and physics faculty out of Europe before the war, undoubtedly delaying the development of Hitler’s bomb. Veblen was also one of the two original faculty (with Einstein) in the Institute for Advanced Study. And Veblen also is remembered fondly on campus for his key stand in 1925 defending an astrophysics professor interested in doing research in the wake of Einstein’s propositions. His victory helped to reshape the institution away from a focus on teaching toward sabbaticals and primary research.
But Veblen’s most important contribution to the topic at hand stemmed from his work and interest in ballistics. In the wake of World War I and the increased mobility of military equipment, much more accurate and timely methods for firing were needed. Veblen undertook the creation of trajectory tables that would take into consideration variables such as altitude, wind, temperature, shell materials, azimuths, and the like to achieve specific firing distances. Each table of 3,000 entries required many multiplications, by hand taking on average 12 hours of error-prone work. Traveling back and forth between Princeton and the Aberdeen Proving Ground, Veblen often thought about how to speed up the work and make the calculation process more efficient.*
In 1930, Veblen invited perhaps the world’s greatest mathematician, 27-year-old John von Neumann, to Princeton University as a lecturer in quantum statistics. Von Neumann became a full professor just a year later. He had received his doctorate in mathematics at age 22 from the University of Budapest and had already published five papers. Three set out a mathematical framework for quantum theory, a fourth was a pioneering effort in game theory, and the fifth explored the link between formal logic systems and the limits of mathematics.
“One of von Neumann’s most remarkable capabilities was his power of instant recall. As far as I could tell, he was able on once reading a book or article to quote it back verbatim; moreover, he could do it years later without hesitation.... on one occasion, I tested his ability by asking him to tell me how A Tale of Two Cities started, whereupon, without any pause, he immediately began to recite the first chapter. We asked him to stop after ten to fifteen minutes.” (Goldstine, p.167)
“von Neumann usually chose a square area roughly two feet on a side on an enormous blackboard and seemed to play the game of seeing if he could confine all his writing to this tiny area...” (Goldstine, p.176)
The Institute for Advanced Study
In 1930, Veblen also helped to organize the Institute for Advanced Study [IAS] in Princeton in part as an escape destination for European mathematicians and physicists. In 1932, Veblen resigned his Princeton professorship to become the first professor at the Institute, where he stayed until he was made emeritus in 1950. John von Neumann joined the IAS in 1933. He was the youngest of the original six professors of mathematics in the IAS, a position he retained for the remainder of his life.
A luxurious new building, Fine Hall (now Jones Hall), housed both the University’s mathematics faculty and the IAS and closely linked them all with mathematical physicists in Palmer physics laboratory. This academic community from about 1933-1939 was unlike any other in America before or since that time and had important consequences for scholarship. With such prestigious faculty in the department and at the Institute, Princeton was quickly considered the Groningen of the 20th century.
The IAS was funded in 1930 by Louis Bamberger, the founder of Macy’s and Bamberger's department stores. He had pulled his funds out of the market just before the 1929 financial crash. Although his support for the Institute and the mathematics department was undeniably generous, those who stayed at the Institute recalled that all of the sheets and towels came from Bamberger’s.
* Veblen’s papers are contained at the Library of Congress, but the Institute for Advanced Study has preserved his tax returns, They show that Veblen had a continuing relationship with Aberdeen from 1918 through the end of WWII, with a substantial uptick in the work in 1942-43. A letter from Marston Morse, Chairman, Committee on Mathematics, to Frank D. Jewitt, National Academy of Sciences summed up the effort during the war: “The work of Professor Veblen of this committee in adding a group of young mathematicians to the Ballistic Laboratory of the Aberdeen Proving Ground is bearing fruit which is astonishing all those who did not know of the inherent ability of the mathematicians chosen.” Records of the ECP, Veblen Paper Box 1 of 1, From The Shelby White and Leon Levy Archives Center, Institute for Advanced Study, Princeton, NJ, USA.
Turing comes to Princeton
In 1935, while his Cambridge undergraduate thesis was being reviewed, Turing attended M.H.A. Newman’s course on the Foundations of Mathematics. There, Turing confronted three notions being set forth by David Hilbert about the inherent nature of mathematics: Was mathematics complete? Was mathematics consistent? And was mathematics decidable? Hilbert had assumed that the answer to all three questions would be “yes.” At its heart, the question was: “Could any mathematical problem be solved?”
By 1928, Gödel had proven that the answer to the first two questions actually was “no,” that arithmetic could not be proved to be both complete and consistent. But the third question remained open. As Hilbert had put it, was there a definite method, a mechanical process that could be applied to a mathematical statement … to determine if or not it was provable?
While on a run outside Cambridge, Turing lay down in a meadow and began conjuring an automatic machine that might prove the decidability of any mathematical assertion presented to it. But what kind of machine? Such a machine, he imagined, would have an unlimited tape marked off into unit squares. The machine would be able to read each square and operate upon it, thereafter moving the tape. Any problem then might be reducible to a set of instructions. In essence, Turing was conjuring up computer algorithms and indeed, the idea of an all-purpose computer.
Before coming to Princeton, in 1936, Turing had documented his remarkable “Turing machine” in a landmark paper, “On Comparable Numbers, with an application to Entscheidungsproblem.” He completed the first draft of his paper in April 1936.
The paper, which spans the gap from philosophical speculation to highly technical mathematics, contained three sections:
- Defining the idea of “computable number” and of “the computing machine.”
- Introducing the idea of a “universal machine.”
- Employing these ideas to prove that the Entscheidungsproblem is unsolvable.
Turing had discovered that no real machine could solve all mathematical problems. But in the process, he had revealed something of far more lasting importance, that such a universal machine indeed could solve many practical problems.
Nowhere in the paper is the slightest hint of its ultimate importance. The approach was so humble – just another thought experiment in the early 20th century – that early on Turing himself may not at all have imagined its impact in shaping future generations.
As was his wont, Turing had worked in isolation, unaware that at Princeton Professor Alonzo Church ’24 *28 (1903-1995), just nine years Turing’s senior, also had taken on Hilbert. In a first paper, Church and students Stephen Kleene and John Barkley Rosser created lambda-calculus. In a second paper, Church used the new system to support his proof that the Entscheidungsproblem is undecidable for Peano arithmetic. Church presented the first paper in April 1935 before the American Mathematical Society. The second appeared in the Journal of Symbolic Logic just as Turing was finishing the first draft of his paper.
Church had attended Princeton, where he publishing his first paper on the Lorentz transformation. He stayed on at Princeton, earning a Ph.D. in mathematics in three years under Veblen. He served briefly as an instructor at the University of Chicago and then received a two-year National Research Fellowship that permitted him to attend Harvard University (1927–1928), and then University of Göttingen and University of Amsterdam the following year. He taught at Princeton from 1929 to 1967, and at UCLA from 1967 to 1990. Under Church’s leadership during the 1930s, Princeton became a leading center of research in mathematical logic, with emphasis upon the completeness and decidability of logical systems. In 1936 Church extended the famous result of Gödel, then visiting the Institute for Advanced Study, by demonstrating the undecidability of first-order logic (Church’s Theorem).
Church’s publication narrowly predated Turing’s, but the approaches were sufficiently different to warrant the publication of Turing’s paper as well.
More than just a publication, the paper brought Turing to the attention of Church. Newman, who appears to have taken some time to get to Turing’s draft, wrote to Church:
“An offprint which you sent me recently of your paper in which you define ‘calculable numbers’ and show that the Entscheidungsproblem for Hilbert logic is insoluble had a rather painful interest for a young man, AM Turing here who was just about to send in for publication a paper in which he used a definition of ‘computable Numbers’ for the same purpose. His treatment, which consists in describing a machine which will grind out any computable sequence - is rather different from yours, but seems to be of great merit, and I think it of great importance that he should come and work with you next year if that is at all possible.”
Newman added: “… I should mention that Turing’s work is entirely independent; he has been working without any supervision or criticism from anyone. That makes it all the more important that he should come into contact as soon as possible with the leading workers on this line, so that he should not develop into a confirmed solitary.”
It was natural that Turing would come to Princeton where, from 1936 to 1938, he was a graduate student in the Department of Mathematics at Princeton and wrote his dissertation under Church, who appears not to have thought of Turing as one his students.
In 1984, Princeton’s Department of Mathematics undertook an Oral History Project, 42 interviews with surviving faculty and students of the mathematics community in Princeton in the 1930s. The interviews include a lengthy one with Church that contains this interesting exchange with William Aspray on 17 May 17, 1984, at the University of California at Los Angeles.
Aspray: Why don’t we turn to your graduate students for a while. If I remember correctly you had Alfred Foster, Stephen Kleene, and John Barkley Rosser. Did you have other students in the’30s?
Church: None that I remember now. There may have been some, but none of note. There was a gap there until later when Leon Henkin and John Kemeny were there at the same time. There were also Hartley Rogers, Martin Davis, Norman Shapiro, William W. Boone, and (much later) D.J. Collins. My memory is very poor, both as to the names and as to the chronological order, but most of these were later than the ’30s.
Aspray: Did you direct Alan Turing’s thesis?
Church: Well, he was at Princeton, but not only under my supervision, because, of course, he had worked with M.H.A. Newman in England. It was while he was working with Newman that his truly original ideas came out.
Aspray: On effectively computable functions?
Church: Yes. In fact, the definitions of effective calculability and the results on the unsolvable decision problems are essentially the same. These were obtained by me and by Turing almost simultaneously. I think I was the earlier by six months or a year. My paper was delayed in publication, but there is an earlier abstract. Turing did not hear of it until it finally appeared. It was, of course, a great disappointment to him. I don’t know the date at which he first had the result.
Aspray: If you don’t mind, I would like to ask a few more questions about this topic, because it is one of particular interest to me since I wrote my dissertation on Turing. How did you hear about Turing’s work?
Church: Well, Turing heard about mine by seeing the published paper in the American Journal of Mathematics . At the time his own work was substantially ready for publication. It may already have been ready for publication. At any rate he arranged with a British periodical to get it published rapidly, and about six months later his paper appeared. At the same time, I think, Newman in England wrote to me about it.
Aspray: Now didn’t his papers appear in the Journal of Symbolic Logic ?
Church: No, I guess there wasn't any such journal at that time. It appeared in a British journal.
Aspray: Proceedings of the London Mathematical Society .
Church: It is quite likely, yes.
Aspray: That is where it was. Did you know Newman at the time?
Church: Only by correspondence.
Aspray: How was Turing’s visit to Princeton arranged?
Church: At Newman’s suggestion he applied for admission as a grad student.
Aspray: I thought that he had come on a one-year fellowship and then was encouraged to stay on by Dean Eisenhart for a second year as a regular grad student.
Church: Yes, I forgot about him when I was speaking about my own graduate students. Truth is, he was not really mine. He came to Princeton as a grad student and wrote his dissertation there. This was his paper about ordinal logics.
Aspray: Right. Did you have much contact with him while he was writing his paper?
Church: I had a lot of contact with him. I discussed his dissertation with him rather carefully.
Aspray: Can you tell me something about his personality?
Church: I did not have enough contact with him to know. He had the reputation of being a loner and rather odd.
Of course, the same often was said of Church.
“Turing made few friends, in part because he befriended only those he considered his intellectual equal and in part because his mannerisms and eccentricities put people off. He had a stammer and a nervous laugh that made conversation with him nearly impossible. He was a sloppy dresser and often neglected to shave for fear of cutting himself – he would often faint at the sight of blood. When he rode around on his bicycle, he would put on a gas mask in an attempt to control his acute bouts of hay fever. Machines and gadgets entranced him. When the chain of his bicycle developed a habit of falling off after a certain number of revolutions, he rigged up a mechanical counter that would keep track of the revolutions and warn him in time to secure the chain.” (Slater, pp. 14-15)
In early 1937, while at Princeton, Turing saw his solution published in the Proceedings of the London Mathematical Society. Later that year, Turing mapped his work to Church’s, he wrote a paper on lambda calculus and two papers on group theory. Ever humble and inept at self-promotion, a Princeton talk about his Turing machine was sparsely attended. He received only two requests for reprints for his paper. Few of his colleagues seemed patient enough to dig into his complex arguments. And as Turing himself put it in a note home to his mother, solving the Entsheidungsproblem was not so much a big deal when Einstein was just down the hall.
With the war coming, cryptography was also on his mind. He conjured up a secure scheme for key-based coding and began dreaming about testing aspects of Riemann’s hypothesis dealing with the density of prime numbers.
To test these two propositions, he stopped by the University’s machine shop. Telephone and telegraph circuits used relays and switches, but Turing may well have been the first to realize that the logical properties of combinations of switches could simulate binary arithmetic and create a Turing machine.
When Turing earned his doctorate at Princeton in 1938 at age 25, von Neumann offered to make him his research assistant at the Institute. Von Neumann was impressed by the young Englishman, although his personal letter of recommendation for Turing failed to mention the work on the Entsheidungsproblem, an ironic omission given von Neumann’s later interest in computing.
Turing declined and returned to England. He subsequently joined the effort at Bletchley Park to break the German Enigma device with the assistance of Colossus, the first operating computer, as part of the Ultra project. Ironically, had he stayed at Princeton, Britain might well have lost the war.
In its early years, the Institute for Advanced Study shared Fine Hall (now Jones Hall) on the Princeton campus. In 1939, IAS offices moved to a new campus a mile away. In the IAS’s Fuld Hall, the League of Nations occupied the top floor. The main floor had offices for Einstein, Veblen, and von Neumann. A decade later, the computer operations were located in a room in the basement now used by a day-care center.
First generation computers
By the late 1930s, von Neumann had become deeply involved in the problems of supersonic and turbulent flows of fluids and by the beginning of WWII, he had become one of the leading experts on shock and detonation waves. As a result, he joined the Ballistic Research Laboratory (1937), with the Office of Scientific Research and Development (1940), with the Navy Bureau of Ordinance (1941), the Manhattan Project (1943), and the Atomic Energy Commission (1955).
During a chance encounter at the Aberdeen train station, von Neumann interrogated Col. Herman Goldstine, one of Veblen’s mathematicians, and came quickly to realize that computers might assist his many research interests. “Our present analytical methods seem unsuitable for the solution of the important problems arising in connection with the non-linear partial differential equations and, in fact, with virtually all types of non-linear problems of pure mathematics. The truth of this statement is particularly true in the field of fluid dynamics. Only the most elementary problems have been solved analytically in this field.” (On the Principles of Large-Scale Computing Machines).
By 1944, eager to find speedier ways to compute, von Neumann became a regular visitor at the Moore School at the University of Pennsylvania. There, a group of engineers constructed the ENIAC (Electronic Numerical Integrator and Calculator), the first electronic digital computer to use vacuum tubes rather than electromechanical relays.
These first-generation computers were enormous, taking up entire rooms. ENIAC stood nearly 10 feet high, 100 feet long, 3 feet deep, occupied a room about 300 square meters in size, and weighed 30 tons.
ENIAC was very expensive to operate in part because it used considerable electricity and generated a great deal of heat. Its vacuum tubes were very prone to failure, had to be replaced every two to three days, and gave rise to a popular field of academic inquiry, fault-tolerant computing. Punched cards and paper tape provided the machine language input into the computer, and the machines could solve only one problem at a time. Magnetic drums were used for memory. Printouts were used for output.
ENIAC continued in operation until September 1955, when it was disassembled. Parts of it now form an exhibit at the Smithsonian Institution in Washington, D.C.
The main two uses of the machine were to calculate Veblen’s trajectories (reducing the calculations from 12 hours to 10 minutes per table) and in 1945, a classified calculation for Los Alamos involving 1,000,000 punch cards, each representing a point in a hypothetical atomic explosion. The building and use of the machine was a controversial war-time priority, but von Neumann’s interest greatly helped to overcome financial objections.
“Von Neumann was the first person, as far as I am concerned, who understood explicitly that a computer essentially performed logical functions, and that electrical aspects were ancillary. He not only understood this, but also made a precise and detailed study of the functions and mutual interactions of the various parts of the computer. Today this sounds so trite as to be almost unworthy of mention. In 1944, it was a major advance in thinking.” (Goldstine)
By late 1944, von Neumann became involved in the discussion about a new machine, EDVAC (Electronic Discrete Variable Automatic Computer) that offered new approaches to storage capacity, a programmable memory, computing speed, sorting speed, the coding of problems, and circuit design.Von Neumann proposed instruction codes as a way to provide program input, and he tested various proposed systems by writing out coded instructions for specific problems. Two of the people involved with the ENIAC/EDVAC set up a partnership, Electronic Control Co. of Philadelphia, that reorganizes soon thereafter into the Eckert-Mauchley Computer Corporation. They succeeded in getting a contract to build a machine for the Bureau of the Census. That machine, UNIVAC, was started in August 1947 and completed in March 1951. It was the first computer commercially produced for businesses.
In June 1945, von Neumann issued a report on the EDVAC that summarized the architecture for most future computers. Goldstine, who later directed the computer effort at the Institute for Advanced Study, described the report as “the most important document ever written on computing and computers.” It set forth clearly the serial mode of a modern computer and described the five principle organs: a unit capable of performing elementary operations of addition; a central processor for executing instructions; a memory for storing numerical data and instructions; and input and output units for intervening between the machine and humans.
Following the completion of the ENIAC and its use to model atomic explosions, Goldstine and von Neumann searched for the means to carry on the development of computers into the post-war world. Were computers a part of the peaceful scientific world? How could research on them be funded? Where was there a congenial atmosphere to take on the work?
The institute machine
In late 1945, Von Neumann initiated a new project at the Institute for Advanced Study in Princeton to create an electronic computing instrument. As first conceived, the Institute, Princeton University, and the Radio Corporation of America (RCA) would collaborate jointly on the project. The director of the IAS, Frank Aydelotte, said to his board: “I think it is soberly true to say that the existence of such a computer would open up to mathematicians, physicists, and other scholars area of knowledge in the same remarkable way that the two-hundred-inch telescope promises to bring under observation universes which at the present moment are entirely outside the range of any instrument now existing.”
All involved in the effort were aware of the project’s importance. Engineer Julian Bigelow, summed it up: “It was happening here … and we were lucky to be in on it… A tidal wave of computational power was about to break and inundate everything in science and much elsewhere, and things would never be the same afterwards. It would cleanse and solve areas of obscurity and debate that had piled up for decades.”
In the end, RCA did not contribute. Instead, outside funding came primarily from the Ordnance Department, the Atomic Energy Commission, and by the early 1950s the Navy and the Air Force. The government interest? To obtain plans for a prototype so that interested agencies could build computing machines of their own.
The Institute for Advanced Study formed a Committee on the Electronic Computer. In the group was Professor John Tukey, a well-known statistician on the Princeton faculty with close ties to Bell Laboratories. The expectation was that such a machine, if intelligently used, would completely revolutionize computing techniques. Von Neumann, who frequently quizzed the project engineers about their performance expectations, expected that the device would be at the very least 10,000 times faster than the present human computer-and-desk-multiplying machine methods.
In 1946, as a tentative aid to programming, von Neumann and Goldstine proposed a crude set of geometric drawings to indicate in rough fashion the iterative nature of a program. That summer, Goldstine became convinced that this type of flow diagram, as they named it, could be used as a logically complete and precise notation for expressing a mathematical problem.
Goldstine arrived full time at Princeton in March 1946 and began the staffing at the Institute. The first hires included Bigelow, who later became chief engineer, and James Pomerene, an engineer with the Hazeltine Corp. On taking the job, von Neumann handed Bigelow a copy of Turing’s computability paper to read as his first “assignment.”
Very early on, von Neumann cast the project as a multifaceted one that embraced engineering, formal logic, logical design, programming, mathematics, and a significant application. By 1948, numerical meteorology emerged as a key application, and Dr. Jules G. Charney, “a first-rate meteorologist,” became the director of the Meteorology Group.
The electronic computing instrument was built at the IAS between 1946 and 1952. During the first months of the project, a logical-design effort resulted in the basic ideas underlying all modern machines. To build the machine, the team created an experimental room in the basement of the Institute and assembled an adequate supply of tools, parts, and testing instruments. Not all of the components were constructed on site. RCA, for example, built the inner memory component and the “Selectron” tube. Eastman Kodak was involved in the construction of the outer memory. The Bureau of Standards modified a teletype for input-output. The Institute retains detailed records on the suppliers, cost, and inventory of every key part.
“It would appear that we have reached the limits of what is possible to achieve with computer technology, although one should be careful with such statements, as they tend to sound pretty silly in five years.” (John von Neumann, 1949)
The machine had four main organs: an arithmetic unit (the central processor), a memory, a control, and input-output. The entire machine was designed to be fully automatic, easy to code and program, and of maximum reliability throughout. It contained as many checking features as technically feasible, and used relatively few vacuum tubes to limit the size of the machine and to improve reliability. Indeed, the IAS machine contained about 2,000 vacuum tubes, compared with 16,000 on the ENIAC.
As sections of the computer were designed, the team shared detailed reports and specifications with universities, other development centers, and with key governmental and military units. The reports survive to this day and convey the frenetic pace of the work. The reports in hand, comparable machines were constructed at other locations, including Los Alamos, Aberdeen, and the University of Chicago, whose machine was nicknamed the Johniac.
After much debate, the team settled on binary rather than decimal for the device, owing to the greater simplicity and the speed with which the elementary operations could be performed. During the mid-’40s, many argued that decimal was better than binary owing to the problem of converting from one to the other. But the IAS team found that there were unambiguous ways to make such operations very simple. Goldstine and von Neumann programmed the binary conversions which were found to take only about 5 milliseconds.
“At the Institute in 1946, there were no tangible assets relevant to computer development except books, brains, prestige, and high hopes – von Neumann and Goldstine in person.” (Bigelow, p. 293)
The effort had many prominent visitors including Turing himself in 1947.
In 1948, under Bigelow’s direction, Jack Rosenberg designed the adder for the arithmetic. Gerald Estrin wrote what may be the first engineering type of diagnostic code that permitted the team to examine every toggle and gate in the machine for correctness of operation.
The 1,024-word memory for the computer was too small to hold both the program and digital information for a problem. A magnetic drum was added to the Institute computer and placed on a limited test basis in May and June 1953. It proved to be very useful, and early in 1955 the need for still more memory inspired the team to procure a 16,384-word drum from Engineering Research Associates, which later merged with Sperry Rand.
During the testing of the arithmetic unit in 1948, the team tested it against von Neumann himself. As they entered in more and more complicated terms, von Neumann finally erred, proving to their collective satisfaction “the power of matter over mind.” (Bigelow, p.302)
In late 1948 and early 1949, the team modified some teletype equipment in order to load the computer’s memory in about eight minutes. Unfortunately, print-outs took 16 minutes, an intolerable delay because such print-outs of memory were required every time they tried to diagnose a machine malfunction. The team prevailed upon IBM for help. Hewitt Crane, who had just left IBM to join IAS, carried out the electronic work needed to adapt an IBM 514 reproducing punch. The input-output times improved in 1952 from minutes to seconds and made decent operation possible.
By early 1949, the team reported major progress in the construction of the arithmetic organ. The machine required two basic types of these organs: two or three shift registers, and an adder (including subtraction) that would operate in conjunction with the shift registers to form an accumulator. These were built in eight stages, and operated satisfactorily under test conditions. The von Neumann machine required 25 microseconds to locate and read a word, compared to 200 microseconds for the EDVAC.
The computer had a basic vocabulary of 29 instructions. Each order consisted, in general, of 10 binary digits to express a memory location and 10 additional digits to express the specific operation.
In 1950, Turing produced another classic paper, “Computing Machinery and Intelligence,” in which he established the basis for the field of artificial intelligence.
During the spring of 1951, programmers were beginning to try exploratory runs on the machine. Error rates were so low that most of the errors were in the human coding. The main use of the machine was the calculations for the hydrogen bomb. One million punch cards helped to maximize the turbulent flows. In the summer, the Los Alamos group ran their thermodynamic program uninterrupted for 60 days. Ivy Mike, the first successful H-Bomb, was dropped on Nov. 1, 1952, on the Marshall Islands. The calculations for the device had taken approximately six weeks.
By July, the computer was in operation for two shifts each weekday.
There were many purely academic uses. Martin Schwartzchild used the machine to investigate stellar evolution. Jules Charney used it for meteorology and found that the calculations were comparable to the insight and training of a highly skilled human. Notably, he successfully re-predicted a hurricane that hit New Jersey in 1950, a run that involved more than 800,000 multiplications.
The machine was used to investigate the mathematics of evolution and the first traffic flows on freeways. Programmers also reverse-engineered the position of planets and the moon back to 601 B.C. They tested a conjecture of the famous 19th-century mathematician Ernst Kummer. The calculation required approximately 20 million multiplications and took six hours of continuous computing.
Although the machine had been operating for a year, the Institute publicly announced the computer on June 10, 1952. Neighbors to the Institute had complained about the potential noise that such a machine might generate. With the public mollified by the reality of the effort, the Institute pressed to obtain good press, although they struggled for an identity distinct from the University. Nonetheless, more than one press account extolled Princeton University’s accomplishments in computing.
To meet the power requirements of the computer and its associated equipment, a 200-ampere feed was installed from the main building load center to the machine location. A closed- circuit air-cooling system provided clean, low-humidity cooling air to the machine. Air was blown through a floor duct into the base of the computer, rising through it, and exhausting through a ceiling duct, returning through an exhaust-blower air filter and cooling coils to the floor duct again. Two remotely located 7 ½-ton compressors provided a year-round cooling operation.
In early March 1953, von Neumann delivered the Vanuxem lectures at Princeton University, a series of four talks entitled Machines and Organisms. An article in Scientific American by John Kemeny of Dartmouth summarized these lectures which also form a considerable part of von Neumann’s The Computer and the Brain (New Haven, 1958).
In May-June 1953, the team developed and operated on a limited test basis a 2,048-word magnetic drum. Then, in August 1953, the IAS computer was moved to its “permanent” location in the mathematical wing of the Institute. The overall height of the machine was too great to permit passage through doorways, so the team had to separate the memory and the arithmetic units. They took advantage of the partial disassembly to carry out several mechanical and engineering improvements on the memory unit. In addition, the control system, power supplies, and cooling system were reorganized to ease operation and maintenance.
The IAS continued the computer program until 1958. After von Neumann left in early 1954, Goldstine headed up the project with Charney, Pomerene, and Maehly leading meteorology, engineering, and mathematical groups. But the era of academic groups making their own computers was coming to a close. The director of the IAS, Robert Oppenheimer, felt that there was a continuing role for the computer at the Institute, but his views did not prevail.
Beginning in 1955, the team gave priority over all new work to the development of a consistent system of general-purpose routines. Essentially a programming language, FLINT (FLoating point INTerpretative sub-routine) went into operation in late 1955.
Another generation of Church’s students in the 1950s and 1960s, notably Michael Rabin, Hartley Rodgers, and Dana Scott, extended his research to automata, formal languages, and formal semantics. The effort helped to shape the new field of theoretical computer science and gave Church’s lambda calculus new life as the basis for functional programming languages.
In August 1955, von Neumann experienced severe pain in his shoulder. He was soon thereafter diagnosed with bone cancer. In 1956, he received the Albert Einstein Commemorative Award and the Enrico Fermi award. His last public appearance came early in 1956 when, in a wheelchair at the White House, he received the Medal of Freedom from President Eisenhower.
By early 1956, it was decided that no new machine would be built at the IAS. Most of the engineering staff therefore left to pursue development work at other places. It was agreed that the IAS Electronic Computer would remain at the IAS, but be transformed from an experimental project into a tool for the solution of the many computational problems arising in the scientific community.
Oppenheimer offered its use formally to Princeton. Before accepting, University Dean of Research Harry Smyth turned to Forman Acton, now professor emeritus in Princeton’s Department of Computer Science but in the mid-’50s Princeton’s de facto computing expert. Acton estimated that the maintenance cost costs would run to more than $100,000. Smyth assured Acton that Oppenheimer had estimated the costs at no more than $10,000 a year. By the end of the first year of operation, the archive confirms that the cost was indeed $110,000. The University quickly lost interest in maintaining the IAS machine, but the extant correspondence makes clear that the University was coming to understand the need for a computer to support University research. Within a year, they had acquired an I BM 650 and installed it in Gauss House, a yellow Victorian home near the present location of Thomas Sweet Ice Cream.
The IAS machine eventually was separated into its component parts, some of which found their way to the Smithsonian. One small part, what appears to be a voltage regulator, remains in the IAS archive.
Von Neumann died in Washington, D.C., at Walter Reed Hospital in February 1957. The cause was bone cancer, which may have resulted from his attendance at nuclear weapons tests and his long experience at Los Alamos. Said Life Magazine: “His death, like his life’s work, passed almost unnoticed by the public, but scientists throughout the free world regarded it as a tragic loss.” A tribute from President Eisenhower cited the “rare and great gifts of mind” von Neumann had given “for the defense of his adopted land and the cause of freedom.”
All but one of the project’s engineers dispersed in the mid- to late-1950s.Herman Goldstine and James Pomerene joined IBM. Ralph Slutz, who had a Ph.D. in physics from Princeton, went to the Bureau of Standards. Willis Ware went to the Rand Corp. Irv Rabinowitz provided computing support at Princeton. Jules Charney left for MIT. Only Julian Bigelow stayed at the Institute.
In 1972, Goldstine published The Computer from Pascal to von Neumann, a book that contains the best surviving account of the IAS effort. In 1980, Bigelow later contributed an important chapter within N. Metropolis, J. Howlett, and Gian-Carlo Rota’s The History of Computing in the Twentieth Century.
Church was elected to the National Academy of Sciences in 1978. In 1985, he received an honorary degree from Princeton. The following statement was read aloud during the Princeton ceremony:
Over some 40 years of research and teaching, he made Princeton an international center of symbolic logic. In work contributing to what has been termed ‘a fundamental discovery of the mathematicizing power of Homo Sapiens,’ he defined the central question concerning the boundaries of formal reasoning. As longstanding editor and reviewer for his discipline’s journal, he gave critical guidance to its quest for the foundations of mathematics and chronicled its history. Through his students, he set a path that has led from the abstract realm of mathematical logic to the concrete domains of computer science and to new vistas of mathematical power.
For the record
Thorstein Veblen was the uncle of Oswald Veblen, who taught mathematics at Princeton. The relationship of the two was incorrect in a story in the April 4 issue.
Aspray, William, “The Mathematical Reception of the Modern Computer: John von Neumann and the Institute for Advanced Study Computer,” in Phillips, Esther R. (ed), Studies in the History of Mathematics, Vol. 26, Washington D.C. 1980, pp. 166-194.
Aspray, William, John von Neumann and the Origins of Modern Computing. Cambridge: MIT Press, 1990.
Bigelow, Julian, "Computer Development at the Institute for Advanced Study", in N. Metropolis, J. Howlett, and Gian-Carlo Rota, A History of Computing in the Twentieth Century, Academic Press, Inc., New York.1980, pp. 291-310.
Dyson, George B., Barricelli's Universe: Digital Computing in Princeton, 1945-1958, Presentation in the Princeton University Public Lecture Series, October 5, 2004.
Dyson, George B., Darwin Among the Machines: the Evolution of Global Intelligence, Cambridge: Perseus, 1997.
Goldstine, Herman H., The Computer from Pascal to von Neumann, Princeton University Press, Princeton NJ. 1972, pp. 237-342.
Hodges, Andrew Alan Turing: the Enigma , New York: Walker and Company, 2000.
Ritchie, David, The Computer Pioneers, Simon & Shuster, Inc., New York, 1986, Chapter 9.
Slater, Robert, Portraits in Silicon, Cambridge MA: MIT Press, 1987, Chapter 3.
Nancy Stern, From ENIAC to UNIVAC: An Appraisal of the Eckert-Mauchly Computers, Digital Press, Bedford, Mass.. 1981.
Stern, Nancy, “John von Neumann’s Influence on Electronic Digital Computing, 1944-4,” Ann. Hist. Comp., Vol. 2, No. 4 1980, pp. 349-362.
Ulam, S., “Von Neumann: The Interaction of Mathematics and Computing,” in Metropolis, N., J. Howlett, and Gian-Carlo Rota. A History of Computing in the Twentieth Century, Academic Press Inc., New York. 1980 pp. 93-99.
Ulam, S., “John von Neumann, 1903-1957,” Ann. Hist. Comp., Vol. 4, No. 2, April, 1982.
Virginia Tech Department of Computer Science, “The History of Computing,” http://ei.cs.vt.edu/~history.
von Neumann, John and Goldstine, Herman, “On the Principles of Large-Scale Computing Machines,” in Von Neumann, Collected Works, vol. v, p.2.
von Neumann, John, The Computer and the Brain (New Haven, 1958).
The Electronic Computer Project, The Shelby White and Leon Levy Archives Center, Institute for Advanced Study, Princeton, NJ. The records of the IAS Electric Computer Project consist of correspondence, memos, reports, patents, contracts, operational guides, blueprints, punch cards, wiring drawings, diagrams, and logbooks. Their comprehensive Finding Aid is available online at http://library.ias.edu/finding-aids/ecpfa.
The Princeton Mathematics Community in the 1930s: An Oral-History Project http://www.princeton.edu/~mudd/finding_aids/mathoral/pm02.htm
In 2003, I interviewed Forman Acton, who provided support for the machine when the University took ownership in 1957, and Roald Buhler, Director the University’s Computer Center in 1966.
In 2004, I had the opportunity to examine the library and personal holdings of Irv Rabinowitz, one of the original programmers of the Electronic Computing Instrument at the Institute for Advanced Study. The following materials were integrated into the holdings the Institute for Advanced Studies Archive.
First Interim Progress Report on the Physical Realization of an Electronic Computing Instrument, Institute for Advanced Study, Princeton, January 1, 1947.
Second Interim Progress Report on the Physical Realization of an Electronic Computing Instrument, Institute for Advanced Study, Princeton, July 1, 1947.
Third Interim Progress Report on the Physical Realization of an Electronic Computing Instrument, Institute for Advanced Study, Princeton, January 1, 1949.
Fourth Interim Progress Report on the Physical Realization of an Electronic Computing Instrument, Institute for Advanced Study, Princeton, July 1, 1948.
First Progress Report on a Multi-Channel Magnetic Drum Inner Memory for use in Electronic Digital Computing Instruments, Institute for Advanced Study, Princeton, July 1, 1948.
Sixth Interim Progress Report on the Physical Realization of an Electronic Computing Instrument, Institute for Advanced Study, Princeton, September, 1951.
Goldstine, Herman and von Neumann, John, Planning and Coding of Problems for an Electronic Computing Instrument, Part II Volume II, Aberdeen Proving Ground, MD, reprinted September 28, 1951 (originally April 15, 1948 at the Institute for Advanced Study)..
Report on Contract No. 7-ONR-388 Task Order I and Final Report on Contract No. DA-36-034-ORD-1023 , Institute for Advanced Study, Princeton, April, 1954.
Final Report on Contract No. DA-36-034-ORD-1330 , Institute for Advanced Study, Princeton, December, 1954.
Final Report on Contract No. DA-36-034-ORD-1646 Part I (Engineering), Institute for Advanced Study, Princeton, December, 1956.
The Use Compiler, Programming Manual I , (for the Univac Scientific Computer 1103A), October 1, 1956.
Final Report on Contract No. DA-36-034-ORD-1646 Part II (Computer Use) for the period from 1 July 1954 to 31 December, 1956 , Institute for Advanced Study, Princeton, May, 1957.
Monthly Progress Report for Contract No. Nonr 1358-(03) and Nonr 1358-(04) , May and June, 1957.
A Short Manual for Flint II, Princeton University Digital Computer, November 1, 1957.