Punched card methods in scientific computation

Thomas J. Watson Astronomical Computing Bureau, 1940. Eckert, Wallace John (1902-1971). Punched card methods in scientific computation. [New York:] Thomas J. Watson Astronomical Computing Bureau, 1940. Original bright orange cloth. ix, 136pp. Text illustrations. (238 x 163 mm). Ex-library card tipped to front paste-down. Library stamps on title page.

"One of the chief proponents of punched-card computation in science was Wallace J. Eckert (no relation to Pres Eckert), as a professor of astronomy at Columbia University. Eckert had become aware of the possibilities of punched-card calculation through visits to the IBM-Columbia University Statistical Bureau, established in 1928; he was also probably inspired by reports from the American astronomer E.W. Brown, who in 1928-1929 had observed L.J. Comrie's use of a Hollerith tabulating system to calculate the positions of the moon. In 1934 Eckert established the Thomas J. Watson Astronomical Computing Bureau at Columbia, which he equipped with modified punched-card machines supplied by IBM. As Eckert described it in a later paper, the Bureau was "the first scientific computing laboratory where general scientific calculations were performed automatically without any reading or writing of figures" (Eckert 1947, 56).

Over the next six years Eckert and his associates developed ways of using the machines to make mathematical tables and to perform complex astronomical calculations. These methods are described in detail in Eckert's Punched Card Methods in Scientific Computation. "Eckert's book marked the advent of the next major advance [in scientific computation], methods of using machines able to process numbers in machine-readable form, where punched cards could be used for the reading, storing, and output of numerical data. Punched card files of mathematical functions eliminated the manual task of looking up and key-entering the values of functions. . . . The speed of reading and multiplying 8-digit numbers by 5-digit multipliers was 1000 multiplications per hour (some four times faster and more accurately than with key-driven calculating machines)" (McPherson 1984, xi). Pages 108-110 contain a description of Eckert's control switch, which allowed sequences of up to twelve operations to be performed on single cards before the next card was read; this marked one of the first steps away from traditional punched-card methods toward the sequence control practices that would later be used in digital computers. (Origins of Cyberspace 2002, 353).

.