NBS Builds a Computer

by Jim Schooley, SAA History Committee  

By the time World War II ended, scientists were pouring endless hours into the creation of electronic calculators. Working in universities, commercial laboratories, or government installations, these researchers sought to displace forever the tedious and error-prone computations performed with slide rules or mechanical calculators. This writer recalls that his mechanical calculator had as one of its virtues the ability to mimic the cadence of, say, the William Tell overture, while it clanked out a particular multiplication. Entertaining yes, accurate yes and faster than paper and pencil, but nowhere near a method to attack the big problems found in mathematics and science. The search for high-speed computation held enormous promise, both for science and for commerce.

NBS scientists played a major role in the highly competitive effort to develop electronic calculators. In fact, in 1950 an NBS team unveiled the "first fully operational stored-program electronic computer in the United States", a machine called SEAC (Standards Electronic Automatic Computer). This is a fascinating story of NBS achievement that can be traced in several archival papers and history books, some of which are listed at the end of this article.

Halting steps toward the realization of electronic computing occurred as early as 1937 at Iowa State University. A computer known as ENIAC (Electronic Numerical Integrator and Computer) was completed at the University of Pennsylvania in 1946. Lightning fast in its computations, the ENIAC could only be programmed by the manual manipulation of switches and plugs. A half dozen other machines were in various stages of development at the same time. All of these machines used hot, relatively short-lived vacuum tubes in their circuitry. Multitudes of these tubes were required to build the machines, and their tendency to fail in service led to agonizing maintenance problems. Invention of the transistor was still "in the works" at the Bell Telephone Laboratories; indeed the very name "transistor" was not coined until mid-1948.

As the development effort intensified, scientists at many universities, corporations, and government agencies–including NBS–could foresee more clearly the enormous value of electronic computers in handling masses of data and in solving problems that required multiple and lengthy calculations. The Bureau of the Census was one of these, and in 1947 Census personnel approached NBS for technical guidance and developmental support in their effort, through a contract with the builders of the ENIAC, to obtain the use of a computer with internally stored, automatically accessible programs. The Census Bureau hoped to use the computer in the course of the 1950 census. The new class of computers was to be called the UNIVAC.

Under the leadership of director Edward Condon, himself a hard-driving theoretical physicist, NBS scientists had plenty of motivation to help the Census Bureau–they wanted a computer too.

In 1947, two NBS groups already had been working for a year on computer-related projects. These groups were located in the Applied Mathematics division and the Electronics division. In NBS Circular 551, issued in 1955, Samuel Alexander described the situation that led NBS to build the SEAC machine: "NBS agreed to serve as technical agent in monitoring the design, construction, and installation of these two machines [one for the Census Bureau and one for NBS] from two commercial suppliers, while its own laboratories were fully occupied with the component development program. [But] Because of an unanticipated sequence of technological and contractual difficulties, it became evident early in 1948 that the delivery of complete machine installations would be delayed considerably beyond the target dates."  Seeing this impasse, NBS threw a substantial number of its staff into an all-out effort to produce a working computer (Alexander mentions that 33 people eventually were awarded a group citation by the Department of Commerce, with special mention of himself, W.W. Davis, R.D. Elbourn, S. Greenwald, R.C. Haueter, A.L. Leiner, S. Lubkin, C.H.Page, J.L. Pike, R.J. Slutz, and J.R. Sorrells). Within 2 years, the group had produced SEAC, which began its useful life in May, 1950.

Samuel Alexander, Sidney Greenwald, and Ruth Haueter described the construction of SEAC in Circular 551. The 20-page article contains much more information about the machine than is suitable here. However, we do learn from it that the machine and its auxiliary components such as the program input and the printer more than filled one laboratory; that SEAC initially contained 750 vacuum tubes and more than 10,000 germanium diodes; that it performed some 11 different types of operations; that both keyboard and tape reader were used as input mechanisms; and that the output went either to a printer or to a tape punch.

SEAC was an immediate hit. Originally designed as a "stopgap" installation for use only until the commercial units became available, its plan was modified when it became clear that no other computer could be completed for at least two years. It became a "nucleus" computer, of immediate use to several organizations upon its completion in 1950, when it began round-the-clock duty, seven days a week. It proved to be a sturdy worker, continuing to solve mathematical problems for some 10 years even though a substantial portion of its time was given to the incorporation of improvements by the design team. Much was learned about computer architecture and reliability during that time.

In an article published in "A Century of Excellence in Measurements, Standards, and Technology", Russell Kirsch described the first 3 years of SEAC operation, including the creation of a second machine labeled DYSEAC. Kirsch also described some of the varied uses that were successfully attempted with the computers. DYSEAC, created expressly for the US Army, was built in movable trailers so as to facilitate its use to control a variety of military field equipment. Although it utilized the same basic circuit elements as SEAC, it incorporated printed circuits which proved to be quite reliable.

Over its useful life, SEAC was used in several applications that gave good indications of the wide-ranging capabilities of computers. A principal application, of course, was numerical storage and calculation. Examples of this type of operation by SEAC are computation of Loran tables for the Hydrographic Office; calculation of missile trajectories for the military; calculation of the strength of wing components for NACA (the forerunner of NASA); and solution of differential equations for nerve fiber reactions. A second class of task was to serve as an experimental station to test new types of peripheral equipment for compatibility and reliability. From the start, SEAC itself operated reliably 77% of the time, an encouraging figure for its designers. Yet another application pioneered with SEAC involved the scanning of textual material and of images. Kirsch provided an example of progress in the area of image processing from his personal experiment with a photograph of his newborn son, scanned using SEAC during the 1950s and again more recently of his son's daughter with a resolution some 1400 times finer.

In this day of talking dolls, give-away calculators, and automobiles with engines that start at the press of a remote control, it is difficult to visualize a world without electronic computers. NBS/NIST employees can justly be proud of the role that their agency played in making these advances happen.

References consulted in the preparation of this article, all of which are available in the NIST library, include the following: "Computer Development (SEAC and DYSEAC) at the National Bureau of Standards" National Bureau of Standards Circular 551, Issued January 25, 1955. "A Century of Excellence in Measurements, Standards, and Technology", NIST Special Publication 958, David R. Lide, Editor, January 2001. "Measures For Progress: A History of the National Bureau of Standards", by Rexmond C. Cochrane, National Bureau of Standards Miscellaneous Publication 275, 1966.

The NBS/NIST Culture of Excellence series is produced under the auspices of the Standards Alumni Association. The SAA, with offices in the basement of the Administration Building, supports NIST management in a variety of ways, but principally by assistance with historical projects such as oral histories of staff members, biographical files, the portrait gallery of outstanding employees, and the museum. Membership in SAA is open to all present and former employees of NIST. For information, call 301-975-2486.