CHAPTER THREEMilitary RootsWHEN historians examine the origins of the electronic digitalcomputer, they usually give top billing to the pioneering efforts ofthe American scientists J. Presper Eckert and John W. Mauchly,who built their Electronic Numerical Integrator and Computer(ENIAC) during World War II. Eckert and Mauchly are justly andwidely honored as the men whose efforts and risks led to the firstmachines recognizable as modern computers. They also foundedthe first private computer systems company. But historians nowalso recognize a lesser known history of the computer, one whoseroots run deep into the most sensitive and secret corners of amodern military establishment.It was no accident that the military services largely financed thepostwar development of the computer in the 1950s, for computingtechnology had played a pivotal role in the Allied war effort. Themilitary indirectly bankrolled even the Eckert and Mauchlycomputer projects, and these relatively open projects were only thetip of a much larger, and sometimes hidden, technological iceberg.The military computer projects, as well as Eckert and Mauchly'sdaring bet on an as-yet nonexistent commercial market, led to thefirst stirrings of interest in computers in the 1950s by business andindustry. Indeed the first commercial computers were direct copiesor adaptations of machines developed for military users.After the war, the commercial potential ofgovernment-sponsored research became more apparent. The firstgreat political clashes over technology policy in the United Statesoccurred.Origins of the ComputerBy the time war broke out in 1939, engineering and science hadalready made important advances in several seemingly unrelatedareas . Although2930 CREATING THE COMPUTERthe war was the catalyst that drew together these advances andproduced the electronic digital computer, its origins stemmed fromfour scientific and engineering traditions. A centuries-old historyof the development of mechanical calculating machines formed thefirst line of activity. The construction of special purpose machines(differential analyzers) designed to approximate the mathematicalsolution of differential equations used in modeling variousphysical processes was the second and more recent type of effort.The third stream of developments resulted from rapid progress indeveloping new generations of electronic components during thewar. Finally, a new theoretical perspective on the abstractmathematical conception of information and informationprocessing preceded further breakthroughs in technology.By the 1930s applied mathematicians had incorporated recentadvances in electrical technology into mechanical calculators.Mechanical levers and gears were being replaced with faster andmore reliable electrical relays and wheels. In the United States,George R. Stibitz at the Bell Telephone Laboratories constructed aseries of electromechanical calculators in the late 1930s that storedand manipulated numbers internally in binary form.1 Themachines were constructed from ordinary telephone relays, wiredtogether in standard telephone equipment racks. They relied onteletype input and output. Larger and more powerful versions ofthese early relay computers were built for wartime use andcontinued in service well into the 1950s.Closely related to this equipment were sophisticated punchedcard business machines that used electromechanical technology totabulate, sort, add, and compare data punched into paper cards.Herman Hollerith invented such machinery, and it was first usedin the U.S. Census of 1890. Although there were significantcompetitors, inparticular, National Cash Register (NCR) andRemington Rand, International Business Machines (IBM)dominated the U.S. market for commercial punched card machinesin the 1930s . IBM had good relations with the government: theestablishment of the U.S. social security system and the ensuingdemand for punched card machinery to process the massivenumbers of1. See W. H. C. Higgins and others, "Electrical Computers for Fire Control,"in M. D. Fagen, ed., A History of Engineering and Science in the Bell System:National Service in War and Peace (1925-1975) (Murray Hill, N.J.: Bell TelephoneLaboratories, 1978), pp. 166-67; and George R. Stibitz, "Early Computers," in N.Metropolis, J. Howlett, and Gian-Carlo Rota, eds., A History of Computing in theTwentieth Century: A Collection of Essays (Academic Press, 1980), pp. 47983.MILITARY ROOTS 31cards needed to keep the system's records was significant inhelping IBM to weather the depression years.2Technologically, both calculators and punched card equipmentrelied on electromechanical relays, counters, adders, comparators,and sensors. Wiring together the more sophisticated mathematicalcapability of the calculator with the data processing capacity ofpunched card equipment was a logical extension of the art.Columbia University researchers put together this type ofinstallation with support from IBM in the late 1920s and used it forcomplex scientific calculations.3 By the later 1930s individualresearchers had begun to experiment with circuitry using vacuumtube flip-flop switches and specialized counters, and otherelectronic circuits, to perform numerical computations.Numerous efforts were under way in the United States. Inspiredby the work of British physicist C. E. Wynn-Williams, who hadbuilt pioneering high-speed nuclear particle counters usingthyratron tubes (a type of gas-filled electronic valve) in the early1930s, Joseph Desch and Robert Mumma of NCR had begunworking on the application of electronics to arithmetic calculationsin 1938 and 1939.4 Installed in a new research lab established atNCR's Dayton, Ohio, plant, Desch and Mumma had by 1940 built aprototype electronic calculator. At MIT, NCR sponsored a parallelresearch effort to build a Rapid Arithmetical Machine, conceivedby Vannevar Bush and supervised by his associate SamuelCaldwell. Although exotic experimental electronic circuitry wasbuilt and tested, MIT never produced a complete machined AtIowa State University, in 1940, John Vincent Atanasoff built afunctional prototype electronic adder. In 1941 he demonstrated it toa visitor, John W. Mauchly.6 IBM also had a small research effortunder way before2. See Bob O. Evans, "IBM System/360," ComputerMuseum Report, no. 9(Summer _984), p. 9; and William Rodgers, Think (Stein and Day, 1969), p. 108.3. See Rodgers, Think, pp. 133-48; and Charles J. Bashe and others, IBM's Earlycomputers (MIT Press,