Today, the northeast corner of the old Moore School building at the University of Pennsylvania houses a bank of advanced computing workstations maintained by the professional staff of the Computing and Educational Technology Service of Penn's School of Engineering and Applied Science. There, fifty years ago, in a larger room with drab- colored walls and open rafters, stood the first general purpose electronic computer, the Electronic Numerical Integrator And Computer, or ENIAC. It spanned 150 feet in width with twenty banks of flashing lights indicating the results of its computations. ENIAC could add 5,000 numbers or do fourteen 10-digit multiplications in a second--dead slow by present-day standards, but fast compared with the same task performed on a hand calculator. The fastest mechanical relay computers being operated experimentally at Harvard, Bell Laboratories, and elsewhere could do no more than 15 to 50 additions per second, a full two orders of magnitude slower. By showing that electronic computing circuitry could actually work, ENIAC paved the way for the modern computing industry that stands as its great legacy.
ENIAC was by no means the first computer. In 1839, an Englishman Charles Babbage designed and developed the first true mechanical digital computer, which he described as a "difference engine," for solving mathematical problems including simple differential equations. He was assisted in his work by a woman mathematician, Ada Countess Lovelace, a member of the aristocracy and the daughter of Lord Byron. They worked out the mathematics of mechanical computation, which, in turn, led Babbage to design the more ambitious analytical engine. This machine, which was never built, encompassed many principles of computer operation that have been rediscovered with newer machines a full century later.
ENIAC was not the first electronic computing device either. By the early 1930s, physicists were already using radiation counters, which employed vacuum tubes as did the ENIAC, and several laboratories before the Moore School had produced devices known as ring counters, which could count from one to ten. In the later 1930s and early '40s, at least three separate efforts to use electronic circuitry to address the problem of computation were made by John Atanasoff, British Intelligence, and IBM. Between 1937 and 1941, John Atanasoff, who taught physics at Iowa State College and had an interest in the general problem of high-speed computation, set out to design a specialized machine that could solve a complex system of linear equations. The Atanasoff-Berry Computer, developed with substantial contributions from his graduate student Clifford Berry, was close to, if not fully operational by 1941.
By that year, IBM, whose expertise then was in punch-card tabulating equipment, had also designed an electronic multiplier. In the late 1930s IBM began to work with Wallace Eckert (2) of Columbia University to explore how their equipment could be used in various scientific applications. It became clear that an electronic multiplier would greatly speed up the kinds of computations being employed by Eckert. IBM had collaborated with him in designing such a system.
Only British Intelligence's Colossus, a computer built at Bletchly Park around 1942, was a large-scale electronic machine. Atanasoff and IBM were limited by the available funds, whereas Bletchly Park and the Moore School tapped into the immense resources for research and development resulting from the war effort. Though highly innovative, none of these specialized computers, unlike the ENIAC, was designed to carry out general purpose computation but served a specific purpose-- much the way specialized particle detectors are designed by experimental physicists to deal with a specific set of phenomena in high-energy physics. The Colossus, a special-purpose machine developed to decode secret messages, performed onlythe logical, as opposed to arithmetical, operations necessary to defeat the famous German code machine Enigma.
In the case of the Atanasoff-Berry computer, the speed was limited by its choice of an electromechanical means of storing numbers, namely the coefficients representing the system of linear equations. As long as Atanasoff's principal scientific interest remained the particular theoretical physics problems motivating the machine's design, his computer was a novel and sufficient solution for those needs. If he chose to use electronics rather than approaching the problem, as he might have done, by means of a complex system of mechanical relays, this reflected a combination of his interests as well as the effectiveness of electronic circuitry with which, like ENIAC's inventors, he had some familiarity.
Invention is almost always a continuous process, with parallel efforts and simultaneous discoveries the norm rather than the exception. This was true for human powered flight as well as for the invention of the electric light bulb, and James Watson and Francis Crick's discovery of the structure of DNA depended greatly on other theoretical and experimental work. This creative tradition of building on the best of the past was true for ENIAC.
Despite its many innovations, ENIAC lacked certain features considered essential to modern computing systems. Without the ability to store a program in its own memory--a feature known as the stored program concept--ENIAC had to be manually wired to execute a particular program. John Mauchly and J. Presper Eckert of the Moore School with John von Neumann and others contributed to this concept. The first machine to operate with this particular design was the EDSAC computer built in 1949 at the University of Cambridge by Maurice Wilkes. Neither ENIAC nor its successor, the EDVAC, had indexed memory and random access memory, which, some might argue going beyond stored program capability, are essential ingredients of modern computer design. Von Neumann and Herman Goldstine at the Institute of Advanced Studies and a team of researchers at the University of Manchester did the most to develop and formalize early computer architectures. Conditional branching--the "IF" statement in a BASIC or FORTRAN program--was not part of the ENIAC's original design.
The Moore School computer nonetheless provided a crucial step in a progression of technological advances; it also served to convince military scientists and technical experts of the value and practicability of electronic computation. The resulting enthusiasm was compounded by the advent of the Cold War; use of electronic computers in the development of the hydrogen bomb laid the foundations for the subsequent computing and information processing industry that has transformed the world since World War II.
John Mauchly's interest in calculating machines was associated with a dream he had of solving "the problem of the weather," an interest his father had shared during his lifetime working on similar problems at Carnegie Institute of Washington. Meteorological research necessitated the computation of enormous amounts of statistical data, and Mauchly, a physics professor at Ursinus College, was constantly looking for ways to achieve more and faster computations than were possible using mechanical desk calculators. At Ursinus, he had already investigated the possibility of using cold cathode tubes. Although these were very much slower than the higher-powered vacuum tubes and had a much more limited margin of operation, they were less expensive and consumed a far smaller amount of electrical energy, making projects more manageable. Mauchly's work on digital electronic circuitry was not sufficiently developed to help him in his meteorological research, and for this purpose he built a somewhat more familiar analog device that he named a "harmonic analyzer."
Mauchly still considered himself a meteorologist rather than a specialist in computational devices. But he was also all too aware that most meteorologists considered his theories "crackpot notions." In fact, he had been so sure these colleagues would not take the statistical results of the work with his "harmonic analyzer" seriously, that, when the American Association for the Advancement of Science met at the University of Pennsylvania in 1940, he delivered a paper on weather statistics to the physics section. One of those in the audience who took a clear interest in Mauchly's talk was John Atanasoff.
The two researchers spent considerable time discussing their mutual interests both then and on subsequent occasions. Though the machine developed by Atanasoff and Berry was a special-purpose computer, it used all-electronic circuitry to perform the addition and multiplication operations at the heart of modern computing equipment. It has received much attention because Mauchly visited Atanasoff and Berry in the summer of 1941 to look at their computer with its electronic calculating circuitry. The use of an electromechanical device to store data and the intermediate results of computation limited its overall speed. Mauchly, who was interested in high-speed, general-purpose computation, reflected his own somewhat different vision for the development of modern computing circuitry when he commented on the relatively slow speed of the design. Seeing the Atanasoff-Berry machine may have encouraged Mauchley by indicating that a larger general-purpose electronic computing machine might be a possibility.
When the U.S. went to war in 1941, many of the Moore School's faculty were called away on secret military research projects or active service. With many new demands for military and communications technology, the War Department sponsored courses of training in the operation of increasingly complicated weapons systems. At the Moore School, a program in Engineering, Science, and Management War Training (ESMWT) was underwritten by the government. Mauchly came to Penn to learn about the latest electronic devices and techniques, and he and Arthur Burks, another Ph.D. enrolled in the ESMWT program, were promptly hired to replace Penn professors who had been called up for active duty.
Meanwhile, the brightest graduate student in the Moore School at the time--described as "undoubtedly the best electronics engineer in the Moore School"--was J. Presper Eckert, Jr. Still in his early twenties, Eckert had already developed an electronic device for measuring magnetic fields and for recording this information on film. The Navy had adopted Eckert's mechanism to assess the performance of their airborne mine- sweeping operations, which were then employing magnetic instruments. Eckert had also already begun his career as an inventor by securing a patent for recording sound on film using diffraction patterns. As an instructor at the Moore School in the summer of 1941, Eckert was hired as an assistant responsible for running the electronics lab associated with the ESMWT course.
The laboratory work of the ESMWT was not much different from what John Mauchly had been teaching his own students at Ursinus, which left plenty of time to chat with Eckert about his major preoccupation, a search for a way to apply electronic techniques to the problems of high-speed computation. Mauchly and Eckert continued their discussions over coffee and fruit sundaes at a local restaurant called Linton's, and in the room containing the Moore School's differential analyzer, the only place with air-conditioning in those early days.
Eckert pronounced Mauchley's ideas on electronic computation to be clearly in the realm of the possible. Using all his engineering creativity, and native genius, he set about addressing the problems that would have to be worked out. Encouraged by Eckert's receptivity to his theoretical notions, and spurred by the serious possibility that his ideas might be implemented, in 1942 Mauchly wrote a five-page memo on the subject entitled "The Use of Vacuum Tube Devices in Calculating." This memo became the basis of the report subsequently submitted by The Moore School to the Army's Ballistic Research Laboratory.
With pressure to produce firing tables for new artillery continually mounting, the need to find faster ways to perform ballistics computations became increasingly urgent. Calculating a trajectory could take up to 40 hours using a desk-top calculator. The same problem took 30 minutes or so on a the Moore School's differential analyzer. But the School had only one such machine, and since each firing table involved hundreds of trajectories it might still take the better part of a month to complete just one table. In his report, Mauchly argued that an electronic machine that could perform 1000 multiplications per second would be able to compute complete ballistics trajectories in minutes rather than days.
When the Allied forces landed in North Africa in 1943, they found themselves operating ordnance in terrain that was entirely different from anything they had previously encountered. The military suddenly needed an entirely revised set of firing tables. With requests growing faster than the calculators could handle them, the backlog of firing tables mounted. This military emergency provided the final impetus for large-scale experimentation in the field of electronic digital computers.
Mauchly's memo was turned into a proposal that could be supported by the Bureau of Ordnance by Captain Herman Goldstine, a mathematician stationed at the Ballistic Research Laboratory (BRL), located at the Aberdeen Proving Ground in Maryland. Goldstine realized that the military was more likely than any other organization to take a calculated risk in time of war. With the approval of John Grist Brainerd, who chaired an important faculty committee, Goldstine presented Mauchly's concept to his superior and arrangements were made for a presentation to the head of the BRL and its chief scientist, Oswald Veblen. On April 2, 1943, a proposal for an "Electronic Diff. Analyzer" was submitted. The name was calculated to forestall anticipated skepticism by associating the proposed computer with the existing differential analyzer. In fact, as a digital device the computer would solve differential equations--the particular mathematical equation used in ballistics problems--by differencing rather than differentiation, the dominant approach at the time. This double entendre was a deliberate subterfuge. More important, the computer described in this report, unlike all previous devices, was to be fully electronic and could compute a ballistic trajectory in under five minutes.
After delivering the initial proposal, Eckert and Mauchly continued to work around the clock to produce supporting arguments and data in anticipation of possible criticisms. On April 9, Eckert's 24th birthday, they presented a more detailed proposal; in May, an agreement was reached; and on June 5, 1942, contract No. W-670-ORD-4926 was signed by the Trustees of the University of Pennsylvania and the U.S. Army Ordnance Department with Brainerd as project supervisor and Eckert as chief engineer. Mauchly was the project's principal consultant, and Goldstine the Army's technical liaison. With a contract now in hand, the machine was named the Electronic Numerical Integrator And Computer, ever after to be known as ENIAC.
In 1942, the first critical problem that had to be solved was construction of a reliable decade counter--an electronic subassembly designed to store and increment numbers from zero to nine. The decade counter was the key component used in a larger unit known as the accumulator, which basically consisted of ten decade counters and their associated control circuitry. Assembled in this fashion, the accumulator could add and store positive and negative numbers from zero to ten billion. Four different types of counters, some based on designs developed elsewhere, were tried out during the first six months of the project.
The major obstacle the inventors faced was the reliability, or otherwise, of the vacuum tubes that were the heart of electronics, which some considered an insuperable problem. Unlike the relatively small number of vacuum tubes used in radios, long-distance telephone systems, and even the more complicated fire control systems (developed by the military for anti-aircraft guns), ENIAC employed vast numbers of these tubes, which could fail unpredictably during long periods of operation. With 17,480 tubes operating at a rate of 100,000 pulses per second, there would be 1.8 billion chances of a failure occurring each and every second. Malfunction of any one of the thousands of tubes, resistors, and condensers could ruin the project. As with any digital calculation, a single failure could alter a number dramatically; one glitch could cause an artillery shell being modeled by ENIAC to suddenly be traveling down instead of up or a hundred times faster than its initial velocity.
Eckert and his team of engineers tested various vacuum tubes, studying when and why they failed in order to eke out a more delicate mode of operation that would increase the life of individual tubes. Lower power levels and careful design alternatives were sought to minimize the amount of work demanded of the vacuum tubes. Most tubes were found to fail early or late in their lives, which resulted in a regimen of preventive maintenance ensuring that only the "healthiest" tubes were used in the ENIAC.
Beyond these careful vacuum tube studies, Eckert instituted rigid requirements for careful design and construction that had to be met by all engineers and technicians on the project where even a faulty soldering joint could render the entire machine useless. Universal design standards, established collaboratively by all of the Moore School engineers ensured that components such as resistors, as well as the vacuum tubes, operated at a certain percentage of their rated capacity. As a result, ENIAC consistently operated for periods greater than the twelve hours Goldstine had proposed as an optimistic estimate.
In his account of these early days, Goldstine called Eckert a "superb engineer:"
Eckert's contribution, taken over the duration of the project, exceeded all others. As chief engineer he was the mainspring of the entire mechanism. Mauchly's great contributions were the initial ideas together with his large knowledge of how in principle to implement many aspects of them. (3)
The ENIAC contract described Mauchly's status as that of principal consultant; as a physicist he was not one of the regular members of the ENIAC engineering team though he understood certain concepts about the uses of high speed computing that others were only gradually beginning to comprehend.
During the entire period of its development, work on ENIAC was enshrouded in secrecy. No papers could be published, and discussion was limited to its initiates. Herman Lukoff, an undergraduate hired during the first summer of the project to design a pulse generator--the unit that generated the "ones" and "zeroes" that were both the data and the control signals for the machine--had no knowledge of its purpose. Lukoff was so excited by the work that he returned as a graduating senior to engineer the power supply for the first two accumulators, a task he completed only hours before he was inducted into the Navy.
In May of 1944, the ENIAC team was able to demonstrate ENIAC's workability in what has come to be known as the two accumulator test. In this, one accumulator was made to increment its value from one to five. The number was then transferred into the second unit one thousand times using the limited control circuitry housed in each accumulator, all in just over one fifth of a second, or about the blink of the eye. At the end of the test, the second accumulator sat idle, displaying the number 5,000--hardly the most impressive of mathematical feats. This demonstration nevertheless caused Dean Harold Pender of the Moore School to express "moderate optimism;" as a veteran in the electrical engineering field, the dean knew the risks but had great faith in his school's engineers. Even so, a great many people continued to doubt that the machine would ever function.
For the inventors, it was a time of elation. They would gladly have continued to devise newer, more clever means of solving problems. But at a certain stage it became necessary to "freeze" the classified design in the interest of completing the task at hand. Nonetheless, as the end of the war approached, engineers at the Moore School began to think intensively about developing a more sophisticated computer. From the first, Mauchly had envisaged the construction of a general-purpose computer, and ENIAC was designed to address this principal concern. Eckert had proposed ways to overcome what he recognized as the major shortcoming of ENIAC, which introduced most of the fundamental elements of hardware design that have become the basis of subsequent computing machinery, with the exception of internally stored instructions. Much of the early discussions focused on the need to increase the machine's memory, for ENIAC could store only twenty numbers in high-speed memory. But the stored program would not be implemented till ENIAC's successor machines were built.
Eckert had experimented with acoustic delay lines early on, drawing on those developed by William Shockley at Bell Laboratories. Now he and his staff investigated the possibility of developing mercury delay lines suitable for use as computer memory. This complex assemblage of mercury tanks, heating units, and electronic control circuitry became the basis of the principal memories in the next generation of computers at the Moore School and elsewhere. These included the EDSAC developed by Maurice Wilkes at Cambridge University (1949); the SEAC developed by Samuel Alexander at the National Bureau of Standards (1950); and the EDVAC, the second large computer built at the Moore School (1951). Eckert and Mauchly also used mercury delay lines in the BINAC (1949) and UNIVAC (1951), two computers they designed and built after leaving the Moore School to establish an independent company.
The idea of storing a program in the same high-speed memory unit as the numerical data, was a major innovation to emerge during ENIAC's construction. It was all too apparent that the manual wiring required to program ENIAC would present an enormous burden that had to be avoided. Before the summer of 1944, Eckert, Mauchly, and other members of the project staff discussed ways of setting up and controlling a computer automatically. In June, 1944, the world renowned mathematician, John von Neumann, joined with the ENIAC team to discuss the formal design of the next generation of computer systems. Most likely drawing upon an earlier version of the stored program concept developed by the famous British mathematician, Alan Turing, von Neumann laid out a formal description of the stored program concept as it might be realized in a high speed computer design. The result of his efforts was "The First Draft Report of the EDVAC Design," a document authored by von Neumann but based at least in part on ideas contributed by others.
In late 1944, the Army Ordnance Department granted a supplemental contract authorizing the Moore School team to begin work on the EDVAC, or Electronic Discrete Variable Automatic Computer.
While many military projects were terminated at the end of the war, ENIAC was not among them. The military's interest in high-speed computing and its use in the nuclear weapons development program ensured the Federal government's continued support of the nascent technology. At the same time, the computer's value for applications far different from problems associated with military weapons and national security came to be recognized by the military and others. A press release issued by the War Department on the occasion of ENIAC's dedication described "The Uses of Computers in Industry," with the computer seen as a means of accelerating economic growth and establishing civilian industries after a devastating war. Commercial uses for computing started to be introduced within a decade of ENIAC's development. Computer technology soon matured into a civilian industry whose growth has been astounding.
Today it is impossible to think of a world without computers or to imagine that the ideas from which these developed and that we take for granted might have been strenuously resisted in the past. The fact is that scientists and administrators involved were skeptical--and with good reason. Running through much that appears in the written record, and explaining some of the discrepancies in the recollections of the many people involved are large doubts as to whether computation by electronic machines would ever be a practical reality. If it seems barely credible today that scientists, engineers, and businessmen five scant decades ago might not at first have grasped the implication of the new technology this has been the case more often than not throughout history, throughout the course of human endeavor. Variations on the theme of "Who needs it?" are followed by the reasons why it can't be done. Examples of early responses to innovations that went on to change the modern world range from Lord Kelvin's observation that radio had no future to Harry M. Warner's skepticism about the market for talking movies. John Logie Baird was considered a lunatic, possibly dangerous, for claiming to have "a machine for seeing by radio." Even in the 1950s, Britain's Astronomer Royal dismissed the notion of space travel as "utter bilge."
During the 1950s, the demands of advanced weapons programs, scientific research and engineering development, and an expanding awareness of data processing applications laid the foundations for a civilian computer industry. The early leaders were the Univac division of Remington Rand Corporation and IBM. Remington Rand (later the Sperry Rand Corporation and now Unisys) acquired the Eckert-Mauchly Computer Corporation in 1950, and was the initial leader in the field. IBM, which introduced the IBM 701 in 1952, gained a predominant position in the computer industry by the mid-1950s, largely through sound product strategies and the efforts of their sales and marketing organizations. Other early manufacturers included Engineering Research Associates (ERA). In 1951, the ERA 1103 computer was actually the first computer system available on the open market. ERA was acquired by Remington Rand in 1951.
Critically important to the sales of early computers were their users. The computing needs of the advanced design efforts of military systems engineering firms, and particularly of the aviation industry in Southern California, provided a major impetus for the growth of IBM's business. The expanding bureaucracies of the federal government--whether in the military logistics and procurement operations of the USAF Air Material Command and the Office of Air Comptroller, or the records keeping of the Census Bureau, U.S. Patent Office, or the Social Security Administration--created a large market for data processing systems. Users also contributed by providing a market for computer systems as well as much-needed technical expertise in the early operation of computer systems. The 1950s saw a severe shortage of scientists and engineers, and computer companies had a hard time financing large programs of computer development. In this situation, volunteer users' associations, such as the IBM users' group SHARE (established in 1956), came up with much of the early applications and systems programming, along with hardware modification recommendations, that IBM's own technical staff was unable to provide.
Solid-state electronics, transistors, and integrated circuits were the revolutionary developments that made possible the miniaturization of computers and heralded the world of today. For a time, development was concentrated in the Northeast, where people like Ken Olsen, the founder of Digital Equipment Corporation where the first minicomputers were produced, contributed to the rapid growth and technical innovations of the industry. Then the action moved out west; the increased use of electronics in the U.S. had created regional pockets of distribution where integrated circuits and other electronic components could be picked up almost as easily as a case of Coca-Cola. Many other people, for whom computer components were as familiar as books, typewriters, or hand calculators, worked as hard as the first generation of computer pioneers to bring about the microcomputer revolution.
ENIAC is one in a long series of innovations that made possible the computing industry of today. Building on past insights and a range of prior work, such as the electronic ring counters designed by RCA, the ENIAC team was also inspired by other individuals, such as the Moore School faculty member, Irven Travis, one of those called up for military service, whose writings on ganged, mechanical adding machines greatly influenced Mauchly's first sketch of the ENIAC's architecture and brought him to Penn for the ESMWT course in the summer of 1941.
ENIAC's design also pointed boldly to the future, incorporating concepts and innovations that went well beyond those developed by earlier researchers and inventors. Regrettably, a dispute over the ENIAC patent soured the memories of many people associated with the ENIAC project and other efforts. The Atanasoff-Berry Computer was judged to be "prior art" by the court in 1973, thereby rendering invalid the ENIAC patent as filed by Eckert and Mauchly. The fiftieth anniversary of modern computing marks a time to recognize the common heritage and the wide range of contributions that so many creative individuals have made to the field of computing.
In planning their public demonstration in 1946, it occurred to Pres Eckert and the rest of the ENIAC team to place translucent spheres-- ping-pong balls cut in half--over the neon bulbs that displayed the values of each of ENIAC's twenty accumulators. Ever since, the flashing lights of computers, often called electronic or giant "Brains" in the early years, have been part of the scene involving computers and science fiction. The development of computers has come a long way since ENIAC's lights first blinked on; but this is only the beginning. The societal transformation brought about by computers world wide has only just begun. The next years may well be even more exciting than those of the past 50 years.
1 As special projects assistant to two Penn presidents, Dr. Winegrad has written articles and two histories about Penn, including Gladly Learn and Gladly Teach (University of Pennsylvania Press 1978), and Through Time, Across Continents (University Museum of Archaeology and Anthropology, 1993). She is now the director/curator of the University's Arthur Ross Gallery, which she helped establish.
Mr. Akera is a doctoral candidate in the Department of History and Sociology of Science at the University.
2 Incidentally, there is no relation to J. Presper Eckert.
3 Herman Goldstine, The Computer from Pascal to von Neumann.Second edition. Princeton University Press, 1993. First edition printed in 1972. 156.
Note: The Adobe Acrobat and HTML versions of A Short History of the Second American Revolution contain changes not available in the print version.
January 30, 1996
Volume 42 Number 18