J. Presper Eckert

Born April 9, 1919, Philadelphia, with John Mauchly , the inventor of the ENIAC, created the EDVAC, BINAC, and Univac computers. Professional Experience: Chief Engineer, ENIAC Project, Moore School, University of Pennsylvania, 1943-1946; founder (with John Mauchly), Electronic Control Corporation (1947, named Eckert-Mauchly Computer Corp.), 1946-1950; director of engineering, Eckert-Mauchly Division, Remington-Rand Corp., 1950-1955; vice president and director of commercial engineering, Remington-Rand Corp., 1955-1959; vice president and executive assistant to the general manager, Remington-Rand Corp., 1959-1963; vice president and technical adviser to the president, Univac Division, Sperry-Rand Corp., 1963-1982. Honors and Awards: IEEE Computer Society Pioneer Award, 1980; member, Information Processing Hall of Fame, Infomart, Dallas, Texas, 1985. J. Presper Eckert was attending the University of Pennsylvania in the Moore School of Electrical Engineering when, as an undergraduate, he became interested in computing. Like so many other students, he was bothered by the fact that his statistics homework required hours of time-consuming use of a calculator to answer his professor's questions. He observed that the school had a copy of the Vannevar Bush differential analyzer and, besides desiring its use, began to think of ways to overcome its limitations. His adviser, Dr. Wygant, proposed to the governmental sponsors that these improvements be implemented, and received a grant which enabled Wygant, Eckert, and a fellow graduate student, Bill Cook, to improve the machine's accuracy and efficiency by an order of magnitude. During this time, John Mauchly See [Mauchly's biography for a discussion of his background and prior interests in digital computing devices.] became interested in the work but decided that there were not many improvements that could be made so long as the calculator was analog. They replaced the gears by installing electronic devices, and devised a mechanism which counted pulses, thus replacing the integrators. However, they felt that even with these improvements the system was still too limited in its applicability. The pulse counter used a monadic representation for numbers and so they decided that they needed a place value system, such as binary or decimal. About this time, the war in North Africa required that all the firing tables be recomputed as a result of the local climatic differences. From this grew their concept of a new machine and with help from John Brainerd, dean of the school, and Herman H. Goldstine, liaison to the school on behalf of the Aberdeen Proving Ground, they presented a proposal for the construction of a digital calculator. There were two main criticisms of the work: mathematicians criticized the use of primitive integration methods in their calculations of the trajectories of ordnance devices, and the engineers predicted that the low reliability of the vacuum tubes would undermine the project. Eckert countered the mathematical objections by pointing out that the intervals were much less than had been used in hand calculations. In any case the flexibility of the system, as a result of the ability to "program" the calculator, permitted the use of more sophisticated methods. There was some concern regarding the cumulative effect of round-off or truncation errors, but this too was overcome. These problems were primarily solved by John Mauchly, who concentrated on the "software" and "programming" problems, while Eckert dealt with the hardware problems, including the tube reliability difficulty. The primary problem that Eckert and Mauchly encountered in this project was the need to use a large number of vacuum tubes (eventually over 18,000), which were notoriously unreliable. [Eckert noted that Fermi heard about their problems and, based on his experiences with equipment including 2-300 tubes, estimated that the mean time to failure of their proposed machine would be 2-3 minutes. Fermi's