Mauchly, John William

views updated

MAUCHLY, JOHN WILLIAM

(b. Cincinnati, Ohio, 30 August 1907; d. Ambler, Pennsylvania, 9 January 1980),

computer science, physics, meteorology, statistics.

Mauchly conceived of and co-invented, with John (J.) Presper Eckert, the Electronic Numerical Integrator and Computer (ENIAC), generally recognized as the world’s first general-purpose, digital electronic computer. The ENIAC was designed and built between March 1943 and November 1945 at the University of Pennsylvania as a wartime project supported by the U.S. Army Ordnance Department. Comprising more than 17,000 vacuum tubes, the machine could perform 5,000 additions per second (and 350 multiplications and approximately 100 divisions or square-root operations per second), making it two to three orders of magnitude faster than the fastest numerical computing machinery then available.

Despite this technical achievement, the historical significance of the ENIAC and Mauchly’s contributions to it lie primarily in establishing the general feasibility of electronic computers and their application to general data processing. Together with Eckert, Mauchly established the Eckert-Mauchly Computer Corporation in December 1947 (which was acquired by Remington Rand in 1950) and built the first commercially successful digital computer, the UNIVAC I (1951), to be sold in the United States.

Family Background and Early Life . John Mauchly was born in 1907 to Sebastian J. and Rachel Scheidemantal Mauchly in the town of Hartwell, Ohio, a suburb of Cincinnati. John’s father, “S. J.” Mauchly, was a high school science teacher who decided to pursue a PhD in physics (degree received in 1913) at the University of Cincinnati. S. J. Mauchly examined the relationship between atmospheric electricity and radio propagation, and he developed an instrument capable of measuring the vertical component of the Earth’s magnetic field. This work and invention earned him a prestigious position at the Carnegie Institution of Washington (Washington, D.C.) and its newly established Department of Terrestrial Magnetism. John’s father then secured his scientific reputation by discovering the diurnal (daily) variation in the Earth’s magnetic field. John was influenced by his father’s reputation and the meticulous attention to data that would guide both individuals’ careers.

S. J. Mauchly’s position at the Carnegie Institution brought the family to the Washington suburb of Chevy Chase, Maryland. Chevy Chase, at the time, was a brand-new streetcar suburb that offered quite a few members of the nation’s emergent scientific elite an opportunity to assert a middle-class identity. Life in the suburbs reinforced John’s aspirations for achievement. John entered McKinley Technical High School in 1921. McKinley was a strong college preparatory school affiliated with the U.S. Navy Yard in Washington. Mauchly excelled in a wide range of subjects. He also became the editor in chief of the school newspaper and earned the rank of commandant in the Washington area High School Cadet Corps.

The other significant influence in John’s early life was the electrical and electromechanical culture of the 1920s. As a youth, John tinkered with his Meccano set, and with electrical devices such as a buzzer and battery. During grade school he devised a block signaling system for his model railroad set. By the time he was in high school, his interest in electrical apparatus afforded him a part-time job repairing electrical appliances for his neighbors and in helping to lay their household lines. In college he also spent a summer as part of an electrical work crew in North Carolina.

Mauchly entered Johns Hopkins University in 1925. Owing partly to his interest in electricity, as well as the requirements of his engineering scholarship, he first chose electrical engineering as his major. However, during his sophomore year he transferred into Hopkins’s doctoral program in physics via a special program for high-achieving students. This decision may have reflected the continued influence of his father. The father also fell prey to a serious ailment—later diagnosed to be bacterial encephalitis—that took his life on Christmas Eve 1928. While still pursuing his doctoral degree, John Mauchly married Mary Augusta Walzl on 30 December 1930; the couple would have two children.

Early Research Involving Calculations . To properly understand Mauchly’s contributions to the field of digital computing, it is important recognize that he began his career as a molecular physicist and not as someone dedicated to machine computation. He began his PhD under Gerhard Dieke, a Dutch émigré physicist known for his work in mass spectroscopy. This work introduced Mauchly to the rigors of calculation. Mass spectroscopy produced a series of photographic traces as intermediate data. It was necessary to then work backward through the physical models to computer molecular energy levels. This was a time when “computers” still referred to men and women who performed computations using calculating machinery. Such work required individuals, including Mauchly, to come up with a meticulous “plan of calculation” that carefully laid out a rigorous procedure for the work.

Mauchly completed his dissertation on “The Third Positive Group of Carbon Monoxide Bands” in 1932. The work was good enough to earn him his PhD; it did not immediately earn him a job. The year 1932 was simply a bad time to graduate. Aside from the Great Depression, U.S. research universities had begun to absorb the initial wave of scientific émigrés. Subatomic physics had also eclipsed molecular physics as the most promising area of study. Mauchly was able to secure a teaching position in 1933 at Ursinus College (Collegeville, Pennsylvania), a small liberal arts college located outside of Philadelphia. This provided him with an income, but neither the time nor resources with which to pursue research. Mauchly attempted for a while to continue his work in molecular physics using borrowed data, but eventually he began seeking other forms of data upon which he could perform analysis.

Mauchly spent the three summers between 1936 and 1938 as a “temporary assistant physicist and computer” at the Department of Terrestrial Magnetism—under his father’s former supervisor, John Fleming. Being at DTM gave Mauchly access to their extensive repository of data on atmospheric electricity. Mauchly was also befriended by Julius Bartels, another scientific émigré, who brought with him a statistical approach to the study of geophysical morphology. Prior to Mauchly’s arrival Bartels had extended Mauchly’s father’s observations by identifying a separate twenty-seven-day cycle in the Earth’s magnetic field. Given that this corresponded to the Sun’s period of rotation, this established an important connection between sunspots, cosmic rays, and the Earth’s magnetic field. Excited by the prospect that his father’s data could be massaged to generate new findings, Mauchly proceeded to comb through the DTM data in search of other regularities.

Based on Bartels’s insights, Mauchly developed a new approach to bivariate statistics. Using this technique, he felt he had identified a diurnal depression in midday ion densities that also exhibited a correlation between angle of incidence (as measured at observatories located at different latitudes) and the intensity of a solar effect. However, the article he submitted to the Journal of Terrestrial Magnetism and Atmospheric Electricity was turned down by Fleming, who served as the editor of the journal. Fleming rejected the article on the grounds that Mauchly lacked the theoretical understanding necessary to draw general conclusions about the cause of magnetic variation in the upper atmosphere. Mauchly also suspected that Fleming had apprehensions about allowing someone outside of DTM to publish an article using the institution’s data.

The subsequent turns in Mauchly’s career go a long way toward explaining the diverse forms of knowledge that he brought to the origins of the digital electronic computer. Fleming’s initial rejection led Mauchly in several related directions. First, it caused Mauchly to shift his interests to meteorology. If the DTM’s atmospheric data had to be regarded as private, the daily weather maps and precipitation data generated by the U.S. Weather Bureau were entirely in the public domain.

Fleming had also challenged Mauchly for relying on too short a period of analysis, and here Mauchly met his supervisor’s challenge through two complementary strategies. The first was to delve further into statistics. A stronger statistical argument could legitimate inferences made from a limited pool of data. Corresponding with the Princeton statistician Samuel Wilks and the Columbia statistician Harold Hotelling enabled Mauchly to develop a more formal, if derivative technique for multivariate statistics called the “sphericity test.” Fleming now accepted Mauchly’s article, published in 1940 as “A Significance Test for Ellipticity in the Harmonic Dial,” though it was clearly rewritten to emphasize the new statistical technique instead of theoretical inferences about the ionosphere. Separately, Mauchly produced a general account of his statistical technique, published in 1940 as “Significance Test for Sphericity of a Normal N-Variate Distribution” in the Annals of Mathematical Statistics, a leading statistical journal. These were the principal technical publications Mauchly issued prior to his work on the ENIAC.

Growing Interest in Computing . A complementary turn was his newfound interest in computing and computing machinery. At Ursinus Mauchly began employing students to work with the weather data, using funds provided by the National Youth Administration, a part of the Works Projects Administration. He also found several occasions to study the latest calculating machinery, including the IBM tabulators on display at the 1939 New York World’s Fair and the Complex Number Calculator developed by George Stibitz at Bell Telephone Laboratories. Working from the published literature, Mauchly developed an analog computing instrument known as a harmonic analyzer (an analog computer uses continuous electrical signals or mechanical motion to directly model physical phenomena), which could identify regularities in statistical data.

As a physicist Mauchly was also exposed to the new “logic” tradition in physics. Electronic “scaling” circuits and “coincidence” circuits emerged as important components within the particle detectors used by atomic physicists. Upon coming across a 1939 article written by E. C. Stephenson and Ivan Getting in Review of Scientific Instruments, Mauchly began to consider how scaling circuits, which in effect could “count,” could be used for high-speed calculating machinery.

Even as he began to explore this idea, his work in numerical meteorology began to bear some fruit. He presented his initial findings during the 1940 meeting of the American Association for the Advancement of Science in Philadelphia. It was during this presentation that Mauchly met John Atanasoff. Through considerable coincidence, Atanasoff was a molecular physicist at Iowa State College who had turned to computational machinery as a means of extending his scientific career. Atanasoff had designed, and, with the aid of his graduate assistant, Clifford Berry, built a specialized computer that used electronics to help solve large systems of linear equations as found in molecular physics and other disciplines.

A main point of historical dispute has to do with the knowledge Mauchly derived from a visit with Atanasoff in Iowa. Although the Atanasoff-Berry Computer was never fully operational, all of the electronic components were assembled and available for demonstration. Mauchly spent several days in Iowa conversing with Atanasoff and his assistant. Atanasoff was also speaking with members of the National Defense Research Committee (the civilian U.S. science mobilization effort) and became aware of the broader use of electronics for computing.

What can be said is that as a result of this visit Mauchly grew excited about the prospects of electronic computers. Given his interest in statistics, Mauchly could envision a very different machine more similar to the mechanical calculators he used for his statistical work. He redoubled his efforts in electronics. He also decided to enroll in a war training program in electronics at the University of Pennsylvania. Then, when the opportunity arose, Mauchly abandoned his tenured position at Ursinus to take up a teaching position in electrical engineering at Penn.

The ENIAC . Mauchly played a crucial role in the development of the the ENIAC (Project PX according to its wartime designation) along with Presper Eckert and other colleagues at the University of Pennsylvania’s Moore School of Electrical Engineering. Physically, the ENIAC was made up of forty frames arranged in a U-shaped layout whose length totaled eighty feet. These frames housed thirty semi-autonomous units, of which twenty were called accumulators, which behaved basically like adding machines in that they could add, subtract, or store a ten-digit number and its sign (+/-). There were two other arithmetic units, namely a high-speed multiplier and a divider/square-rooter. The ENIAC also had three function tables with a large array of manual switches whose values could be read at electronic speeds. These function tables were necessary to represent the nonlinear, incalculable component of the mathematical equation for exterior ballistics (the flight of an artillery shell), which was considered the most important application by ENIAC’s sponsor, the Army Ordnance Department.

Meanwhile, the easiest way to recognize the novel design of the ENIAC is by looking at its programming system. It is significant in this respect that the ENIAC’s accumulators did not have all of the functions of an adding machine. While a mechanical adding machine stores two numbers, x and y, in computing the sum x + y, an accumulator was designed to store only one number so that every addition (and subtraction) had to occur by precisely coordinating the action of two different units. Similar coordination was required to perform both multiplication and division. Coordination was achieved through a system of wires and plug boards that routed a program pulse. The first program pulse was generated by the ENIAC’s initiating unit. And whereas there was a master programmer used to execute a particular operation for a fixed number of iterations, programming the ENIAC required the careful coordination of events at the scale of the entire machine. (Though not part of its original design, the ENIAC could also execute a conditional branch instruction through a procedure known as magnitude discrimination, which tested for the sign of a given number. This was a later addition, not made by Mauchly.)

While this description is a way to visualize the ENIAC as it was designed, it remains an anachronistic description that obscures the design’s historical origins. Historically the ENIAC and Mauchly’s contributions to it emerged through the confluence of several related developments. The first was the development of an analog mechanical computing instrument known as the differential analyzer, and the presence of one of these units at the Moore School. The differential analyzer was developed by Harold Hazen and Vannevar Bush at MIT, an extension of which was made by Moore School instructor Irven Travis under contract with the Army Ordnance Department.

Upon completing this work, Travis gained recognition as a regional expert on computing instruments, and was hired by General Electric as a consultant. One of the reports Travis wrote for GE described the possibility of a numerical alternative to the differential analyzer, one that was based on ganging together a bank of adding machines; he also mentioned the possibility that electronics might be necessary to deliver the requisite speed. Mauchly insisted that he never saw a copy of Travis’s report. However, Mauchly did speak with Travis at some length about computing machines prior to Travis’s departure for a war assignment.

Finally, war mobilization brought the Army Ordnance Department, and its new Ballistic Research Laboratory (Aberdeen, Maryland), to renew its affiliation with the Moore School. In addition to commandeering the Moore School’s differential analyzer, BRL asked the Moore School to assemble a human computing unit, and to use numerical methods to produce ballistics tables. That such work was taking place at the Moore School was well known; Mauchly’s wife, Mary, also served as one of the human computers prior to the Army’s evaluation of the ENIAC proposal.

For a while, Mauchly’s interests remained with meteorology. However, as his interest in electronic computers grew, the Army’s need for ballistics tables—and the considerable backlog that existed by late 1942—provided a suitable rationale for building one. Though the evidence here is scarce, it appears that Mauchly did consider the possibility of designing the numerical equivalent of a differential analyzer using the circuit design techniques employed by Atanasoff (Atanasoff basically used coincidence, or real “logic” circuits rather than counters, to perform arithmetic). Kindly rebuffed by Atanasoff based on the advice Atanasoff received from his patent attorney, Mauchly turned instead to the idea mentioned, if not sketched out by, Travis. Though the idea was not fully original, it at least “belonged” to the Moore School.

The ENIAC’s design was clearly influenced by the attempt to implement a numerical alternative for the differential analyzer. What one sees as its “data-flow architecture” owes its origins to the differential analyzer (and to analog computers more generally), where “data” travel simultaneously throughout the system during computation. However, numerical methods were substantially different from those of analog computation. Computation had to take place through sequential operations that spread the computation out over time. This was something entirely familiar to the human computers who produced ballistics tables; it was also something familiar to Mauchly through his knowledge about creating a “plan of calculation.” Moreover, so long as the ENIAC was designed to support numerical calculation, it was not limited to a specific class of mathematical problems, as was the differential analyzer. The ENIAC, in any event, was an amalgam of these two technical traditions.

The fact that wartime exigencies led to a rushed design effort is evident in both versions of the ENIAC “proposal.” The first was an informal memo submitted by Mauchly to the Moore School’s war projects supervisor in August 1942; the second was the formal proposal Mauchly and Eckert submitted to the army in April 1943 with the assistance of at least one other engineer. Though both documents spoke of a central programming device, neither had any substantial description of how such a unit would allow the ENIAC to execute a defined sequence of operations. There was still no design for the programming system when the army officially approved the project in June 1943.

The ENIAC’s programming system emerged in a rather contingent manner. Following standard engineering practice, the project engineers set out to demonstrate a proof of concept by assembling two accumulators. It was necessary to interconnect the two units to provide a meaningful demonstration. And despite mention of a central program control unit, it was easier to implement the interconnection through a local system of controls placed on each accumulator. Hoping, nevertheless, not to have to substantially reengineer the accumulator as they moved beyond the two-accumulator test, those on the project, including Mauchly, made an educated guess about the number of local controls necessary for future use. Most likely, the decision to take such an approach was reinforced by the fact that the differential analyzer, in its normal mode of operation, was physically reconfigured for each equation. This was the historical origin of the distributed programming system that became the hallmark of the ENIAC.

The historical literature has been somewhat divided about Mauchly’s contributions to the actual development of the ENIAC. The most important thing to note is that Mauchly served as a consultant, rather than as the project’s director. Project PX was a major wartime project, one that the university did not entrust to an untenured faculty member with so little background in electronics. This is not to say that Mauchly had no role in the project. Mauchly kept an eye on the validity of the mathematical algorithms, was among those who reviewed the programming system, and was actively involved with filing patent disclosures. (Aside from his interests in doing so, this was made necessary by the university’s contractual obligations to the military.) Mauchly also served as an important sounding board for Eckert on all aspects of the ENIAC’s design. Yet, so long as Project PX was cast as an engineering endeavor, where the Moore School was expected to deliver the machine described in the proposal, the principal expertise required for doing so lay with the Moore School engineers. As the project wore on, Mauchly, at least in the eyes of the engineers, became more an adviser to the project than someone directly involved in the crucial task of building the machine.

However, existing accounts of Mauchly’s contributions are based on a misconception of the process of invention. Moreover, the historical literature tends to lump together Mauchly and Eckert’s contributions and describe them as sole inventors, which is itself an artifice of the system of assignments in the U.S. patent system. Mauchly’s contributions, and those of his colleagues at the University of Pennsylvania, were incremental in nature, in a way that is consistent with what is known in general about the inventive process.

It should also still be clear that Mauchly played a crucial role in the invention of the ENIAC. Despite all of the resources that were devoted to war research, Mauchly alone possessed the different forms of knowledge and technical interests, and resided at an institutional location that produced a versatile, digital electronic computer during World War II. Other technical developments in electronics and the considerable advances in applied mathematics during the war would have ensured that a digital electronic computer was not far off on the horizon. Nevertheless, Mauchly stood at the crucial juncture necessary to “invent” the ENIAC.

Subsequent Career . Shortly after the war, and the successful completion of the ENIAC, Mauchly suffered a personal tragedy when his wife, Mary, drowned in an accident in 1946. In 1948, Mauchly married Kathleen “Kay” McNulty, a mathematician who had been one of the first ENIAC programmers. After the war, Mauchly also found an opportunity to try his hand at entrepreneur-ship. The cultural cachet of being an “inventor” was not something so easily cast aside. Mauchly, along with Eckert, established the Electronic Controls Company in 1946, and the firm was formally incorporated as the Eckert-Mauchly Computer Corporation in December 1947. Working first with a contract from the U.S. Census Bureau (the contract was also supervised by the National Bureau of Standards), Eckert and Mauchly began cultivating a market for their digital computer, the UNIVAC I. At EMCC, and then at Remington Rand, Mauchly focused his energies on the study of computer applications and on marketing. He was among those who helped develop a computer’s early instruction sets (the operations that any given computer directly recognizes and supports), by working with the early programmers at the National Bureau of Standards.

Mauchly’s self-taught marketing skills were eventually eclipsed by the professionally trained marketing staff of

Remington Rand. In 1959 he left the firm to create his own independent consulting firm, Mauchly Associates, which he then followed with another venture, Dynatrend, in 1967. Here he found a more permanent niche as a consultant within the emerging computer services industry, where he helped to introduce quantitative project management techniques to U.S. construction firms and other businesses.

Mauchly and Eckert would have retained their reputation as the inventors of the digital electronic computer had it not been for an acrimonious patent dispute (Honeywell v. Sperry Rand Corp), filed with the federal district court in Minnesota, whose 1973 decision invalidated the 1964 ENIAC patent. Debates from this latter period have polarized historical memories, and have tended to occlude Mauchly’s contributions. Mauchly retired to the quiet suburb of Ambler, Pennsylvania, outside of Philadelphia. He died in 1980, at age seventy-two following several bouts with illness. His wife, Kathleen Mauchly, survived him, marrying the photographer Severo Antonelli in 1985; Kathleen McNulty Mauchly Antonelli died in April 2006.

BIBLIOGRAPHY

John Mauchly’s papers may be found in the Annenberg Rare Book and Manuscript Library, University of Pennsylvania, Philadelphia, Pennsylvania. See also UPD 8.10, ENIAC Papers (ENIAC Patent Trial Collection, 1864-1973), University Archives and Records Center, University of Pennsylvania.

WORKS BY MAUCHLY

“A Significance Test for Ellipticity in the Harmonic Dial.” Journal of Terrestrial Magnetism and Atmospheric Electricity 45 (September 1940): 145–148.

“Significance Test for Sphericity of a Normal N-Variate Distribution.” Annals of Mathematical Statistics 11 (1940): 204–209.

OTHER SOURCES

Akera, Atsushi. Calculating a Natural World: Scientists, Engineers, and Computers during the Rise of U.S. Cold War Research. Cambridge, MA: MIT Press, 2006.

Aspray, William. John von Neumann and the Origins of Modern Computing. Cambridge, MA: MIT Press, 1990.

Burks, Alice R., and Arthur W. Burks. “The ENIAC: First General Purpose Electronic Computer.” Annals of the History of Computing 3 (1981): 310–389. Best article-length technical description of the ENIAC.

_____, and Arthur W. Burks. The First Electronic Computer: The Atanasoff Story. Ann Arbor: University of Michigan Press, 1988.

Costello, John. “As the Twig Is Bent: The Early Life of John Mauchly.” IEEE Annals of the History of Computing 18 (1996): 45–50.

Goldstine, Herman H. The Computer: From Pascal to von Neumann. Princeton, NJ: Princeton University Press, 1993. A historical account published by one of the participants. Originally issued in 1972.

McCartney, Scott. ENIAC: The Triumphs and Tragedies of the World’s First Computer. New York: Walker, 1999. An accessible, nonacademic history of Mauchly and Eckert.

Stern, Nancy. From ENIAC to UNIVAC: An Appraisal of the Eckert-Mauchly Computers. Bedford, MA: Digital Press, 1981.

Atsushi Akera