The application of commercial calculating machines to scientific computing
L. J. Comrie · Mathematics of Computation · 1946
The Application of Commercial CalculatingMachines to Scientific Computing I propose to deal with one era in computing, but will first recall some of the landmarks in that art.Perhaps the first was the introduction of the decimal system.Then came the invention or discovery of logarithms in 1614 by Napier, which enabled multiplications and divisions to be done much more rapidly than had been the case before.Briggs' industry in tabulating logarithms of numbers and of trigonometrical functions made Napier's discovery immediately available to all computers.Although Pascal constructed the first calculating machine in 1642, and although Leibniz, Morland, Stanhope and others contributed ideas and designs, nearly two centuries were to elapse before a commercial calculating machine could be purchased.This was the arithmometer of Thomas de Colmar, first used in actuarial offices in Paris in the 1820's.It uses the stopped-wheel principle enunciated by Leibnitz to determine which digit shall be added in each columnar position.There is little to report till about half a century later, when Odhner invented his pin-wheel for conveying varying digits to the machine.This principle is still used in the Swedish Odhner and other barrel-type crank-driven hand machines, particularly the German Brunsviga, which acquired manufacturing rights about 1890.This marked the beginning of a period of development, which continued with ever-growing intensity till, for the second time this century, progress was interrupted by war.These developments include keyboard setting, electric drive, automatic control of multiplication and division, duplication of registers, and storage and transfer facilities.Although these machines were developed primarily for the wide fields of commercial and accounting application, they lend themselves to mathematical and scientific computing, which consists, after all, of the four fundamental processes of addition, subtraction, multiplication and division -just the processes that any calculating machine is designed to perform.Between the two world wars comes the era under review-that in which scientists discovered more and more how numerical processes could be lightened by mass-produced commercial machines.To make such a discovery, it is necessary that one person should understand both the machine and the problem.As far as possible commercial machines should be used in their standard form, although minor alterations are permissible.The computer's art lies mainly in his ability to manipulate the problem and to apply ingenuity and low cunning in developing techniques that take advantage of the mechanical features of the machine.If I express the view that this process of adaptation has gone further in Great Britain than in U. S. A., I should like to give a rational explanation of why this should be so.Our great advantage is that we have more material with which to operate.In U. S. A. there are only three outstanding crankdriven calculating machines-the Marchant, Monroe and Friden.These are similar in capacity and in possessing keyboard setting, electric drive, automatic multiplication and division, but few special features such as one finds in the German Brunsviga, Mercedes, Archimedes and Rheinmetall, or the