Skip to Main Content
Have library access? Log in through your library
The Computer from Pascal to von Neumann

The Computer from Pascal to von Neumann

Herman H. Goldstine
Copyright Date: 1993
Pages: 365
  • Cite this Item
  • Book Info
    The Computer from Pascal to von Neumann
    Book Description:

    In 1942, Lt. Herman H. Goldstine, a former mathematics professor, was stationed at the Moore School of Electrical Engineering at the University of Pennsylvania. It was there that he assisted in the creation of the ENIAC, the first electronic digital computer. The ENIAC was operational in 1945, but plans for a new computer were already underway. The principal source of ideas for the new computer was John von Neumann, who became Goldstine's chief collaborator. Together they developed EDVAC, successor to ENIAC. After World War II, at the Institute for Advanced Study, they built what was to become the prototype of the present-day computer. Herman Goldstine writes as both historian and scientist in this first examination of the development of computing machinery, from the seventeenth century through the early 1950s. His personal involvement lends a special authenticity to his narrative, as he sprinkles anecdotes and stories liberally through his text.

    eISBN: 978-1-4008-2013-9
    Subjects: History of Science & Technology

Table of Contents

  1. Front Matter
    (pp. i-iv)
  2. Table of Contents
    (pp. v-viii)
  3. Preface (1993)
    (pp. ix-x)
  4. Preface
    (pp. xi-xii)
  5. PART ONE: The Historical Background up to World War II

    • Chapter 1. Beginnings
      (pp. 3-9)

      There is of course never an initial point for any history prior to which nothing of relevance happened and subsequent to which it did. It seems to be the nature of man’s intellectual activity that in most fields one can always find by sufficiently diligent search a more or less unending regression back in time of early efforts to study a problem or at least to give it some very tentative dimensions. So it is with our field.

      Since this is the case, I have chosen somewhat arbitrarily to make only passing references to the history of computers prior to...

    • Chapter 2. Charles Babbage and His Analytical Engine
      (pp. 10-26)

      The theme of Leibniz—to free men from slavery by the automation of dull but simple tasks—was next taken up by one of the most unusual figures in modern intellectual history, Charles Babbage (1791-1871). As his biographer has so correctly said of Babbage everything about him was contentious including the date of his birth. According to her, he was born on 26 December 1791, in Devonshire, even though he stated that it was in 1792 in London. The title of her bookIrascible Geniusis another good clue to his character.¹

      In any case, he was born into an...

    • Chapter 3. The Astronomical Ephemeris
      (pp. 27-30)

      Until about the middle of the eighteenth century there was no satisfactory method available to a mariner for calculating his longitude. This, of course, constituted an extremely serious problem for sailors to navigate over great distances and arrive at predetermined ports with any accuracy.

      In principle, it is not difficult to determine longitude given anaccuratetable of the moon’s position as a function of the time. The problem lies in making such tables, because the moon’s motion is very complex: it is compounded principally out of the mutual interactions of the moon with both the earth and the sun....

    • Chapter 4. The Universities: Maxwell and Boole
      (pp. 31-38)

      Thus we observe in Babbage’s period a great flowering of astronomy and celestial mechanics. It was still too soon however to see a similar situation obtain in other branches of natural philosophy. As a consequence, great calculations were not attempted or, indeed, needed in other branches of physics. In fact, we may say as a general principle that large calculations will not be attempted in a given field until its practitioners can write down in mathematical form an unambiguous description of the phenomena in question.

      This was not possible in Babbage’s time, for example, in the field of electricity, where...

    • Chapter 5. Integrators and Planimeters
      (pp. 39-51)

      To describe the next major advance in the field of computing and to set the stage for our real topic, the early history of the electronic digital computer, we need to embark on an excursus. It is desirable at this point to acquaint those of our readers who are not already familiar with them with the differences between digital and analog or continuous computing instruments. These are the two broad categories into which we can divide all forms of computing equipment or processes.

      Every computing instrument or machine of whatever sort has certain fundamental operations which it performs and out...

    • Chapter 6. Michelson, Fourier Coefficients, and the Gibbs Phenomenon
      (pp. 52-59)

      There was one substantial difficulty with Kelvin’s Harmonic Analyzer: it was not possible to increase by very much the number of terms in the series since his device for adding the terms together led to an accumulation of errors. If the number of terms was fairly large, this accumulation could vitiate the results completely. In fact, Albert A. Michelson (1852–1931) and Samuel W. Stratton (1861–1931) were to write in 1898:

      The principal difficulty in the realization of such a machine lies in the accumulation of errors involved in the process of addition. The only practical instrument which has...

    • Chapter 7. Boolean Algebra: x² = xx = x
      (pp. 60-64)

      Earlier we discussed very briefly George Boole’s contribution to logics and hinted at what he had done. Perhaps now it is worth our while to describe in somewhat more detail the nature of Boolean algebra, since we can do this in not too technical terms and since it has such relevance to digital computers. As Boole said, his aim was to “give expression ... to the fundamental laws of reasoning in the symbolic language of a Calculus.” He did this by a most ingenious reduction of logic to algebraic form. It is worth our while to try to understand a...

    • Chapter 8. Billings, Hollerith, and the Census
      (pp. 65-71)

      At this point we take up consideration of digital machines that are basically founded on Boole’s ideas. Here we find an anomalous development. Up till now we have dealt with scientific and engineering usages and needs for computation. Now all of a sudden we find a totally new situation confronting us: needs of the Census Office of the United States Department of Interior are such in the 1880s that a system of machines is developed to automate at least partially the procedure. We have here then for the first time a statistical and a commercial need for computation. A very...

    • Chapter 9. Ballistics and the Rise of the Great Mathematicians
      (pp. 72-83)

      Let us now put to one side the development of punch card equipment and return to discussions of scientific and engineering needs and advancements in the field of computing. In particular, let us examine what World War I did in these respects.

      In a one-sentence oversimplification, that war was, from a scientific point of view, dominated by chemistry and did much to further the field in the United States. At the beginning, the United States was extremely dependent on Europe for potash, nitrates, and dyes. To overcome this dependence the government did a number of things, including the construction of...

    • Chapter 10. Bush’s Differential Analyzer and Other Analog Devices
      (pp. 84-105)

      What we wish now to do is to set the stage for the works of Vannevar Bush and his colleagues at MIT; of Howard Aiken at Harvard and his colleagues at IBM; and of George Stibitz and his associates at the Bell Telephone Laboratories. To do this we need to gain a little understanding of the degree of mathematical sophistication of the electrical engineer in the 1920s.

      Ever since early days there has been a fairly close linkage between various branches of engineering and the cognate sciences. In many cases of course the links were formed by natural philosophers or...

    • Chapter 11. Adaptation to Scientific Needs
      (pp. 106-114)

      We have now finished with our main discussion of analog computers and wish to turn back to digital ones. In doing this we have purposely omitted mention of a variety of analog machines of great ingenuity such as Mallock’s machine for solving systems of linear equations. The interested reader is again referred to the works of Murray and d’Ocagne.¹

      It is perhaps worth remarking that a Spaniard, Torres Quevedo (1852-1936) proposed in 1893 an electromechanical solution to the ideas of Babbage. His machine may be viewed as an extremely tentative step in the chain of development of digital devices. It...

    • Chapter 12. Renascence and Triumph of Digital Means of Computation
      (pp. 115-120)

      Concurrent with Aiken’s collaboration with Lake and his associates, George R. Stibiz of the Bell Telephone Laboratories, with the help of two most able engineers, Samuel B. Williams and later Ernest G. Andrews, also of those laboratories, was busily engaged on a similar task. Both groups started in 1937 and Stibitz’s team produced its first device in 1940; this “partially automatic computer” was as shown to the American Mathematical Society at a fall meeting at Dartmouth College, where appropriately so many important things have been done in the computer field—including the appointment of John Kemeny, the inspiration for much...

  6. Illustrations
    (pp. None)
  7. PART TWO: Wartime Developments:: ENIAC and EDVAC

    • Chapter 1. Electronic Efforts prior to the ENIAC
      (pp. 123-126)

      As we saw in the last pages of Part One, the electromechanical digital devices were doomed to extinction by the advent of electronics. It is curious how delicate are the timing and balance of what we now refer to as research and development. There seems to be an optimal time for discovery, as well as an optimal period for the perfection of the idea. Thus, if one is late in starting or dilatory in carrying out the task, one may, as John Couch Adams almost did in connection with the discovery of Neptune, find one-self “scooped.” If, on the other...

    • Chapter 2. The Ballistic Research Laboratory
      (pp. 127-139)

      Before following up on this story, we need to digress long enough to bring things into their proper temporal sequence. To do this it is convenient first to introduce a number of people whose careers are tied in closely with our story.

      In 1910 Hermann H. Zornig graduated from Iowa State College and was commissioned in the regular Army of the United States that summer. He subsequently did graduate work at the Massachusetts Institute of Technology and the Technische Hochschule in Charlottenberg, Germany, where he had the unique opportunity to study under one of the greatest figures in ballistics, Carl...

    • Chapter 3. Differences between Analog and Digital Machines
      (pp. 140-147)

      We noted earlier that the nineteenth-century physicists had attained an ability to describe in mathematical form quite complex machinery and were therefore able to invert this process. Given a mathematical formula, they were, in principle at least, able to invent a machine exactly describable by the formula. This is what the analog computer is. What are the limitations on this idea? There are fundamentally three aspects to the problem having to do with the generality, the accuracy, and the speed of analog equipment.

      While it is in principle possible to find a machine capable of carrying out a given calculation,...

    • Chapter 4. Beginnings of the ENIAC
      (pp. 148-156)

      The reader should keep constantly in mind that at this time a truly definitive history of the electronic computer cannot be written since many things that happened in the early days (1943–1957) are still controversial and others are perhaps still classified. In setting down this account I have however felt that these dangers are less than the advantages that will accrue to later historians in having the detailed account of one of the few participants in this period who is still alive and in possession of a very complete set of the original documents. As I stated in the...

    • Chapter 5. The ENIAC as a Mathematical Instrument
      (pp. 157-166)

      At this point we ought to pause and describe the ENIAC as a mathematical instrument so that the reader may better understand what it did and how. In doing this we should understand that the machine was unique and primitive; most of the underlying architectural and organizational ideas were abandoned after it was completed, and therefore we are in some sense dissecting a dinosaur.

      A paper written at the time describes the machine generally in these terms:

      The machine is a large U-shaped assemblage of 40 panels . . . which together contain approximately 18,000 vacuum tubes and 1,500 relays....

    • Chapter 6. John von Neumann and the Computer
      (pp. 167-183)

      John Louis Neumann was born into a well-to-do family in Budapest on 28 December 1903, during the last socially brilliant days of that city under the Hapsburgs. His father Max, a banker, was a partner in one of the city’s important private banks and was able to provide well for his children both intellectually and financially. He was ennobled in 1913 by the Emperor with the Hungarian title ofMargattai, which young von Neumann later Germanized tovon. His father and mother, Margaret, had three sons, John, Michael, and Nicholas, of whom John was the eldest.

      While still very young,...

    • Chapter 7. Beyond the ENIAC
      (pp. 184-203)

      To describe what happened next it is convenient to refer back to Babbage’s Analytical Engine, to Stibitz’s relay computers, to the Harvard-IBM computer, and to Post’s and Turing’s paper constructs. We have seen (above, p. 21) that Babbage’s machine was conceived as being instructed in its tasks by a set of so-called operation cards strung together to describe the series of operations to be performed. Similarly, the other machines just mentioned all had paper tapes into which were punched holes that were a numerical code for the instructions to be effected. As early as November of 1943 Stibitz was describing...

    • Chapter 8. The Structure of the EDVAC
      (pp. 204-210)

      We can now describe in some detail the logical structure of the EDVAC. It was the immediate antecedent of the Institute for Advanced Study computer, which was to be the prototype for the systems of today, and is thus important historically. In 1945 however these things were as yet no more than dreams of the future. But von Neumann’sFirst Draftwas to be a blueprint for a whole line of machines starting with the EDSAC at the Cavendish Laboratory in England and continuing through a number of other delay line machines.

      This machine, as we saw earlier in the...

    • Chapter 9. The Spread of Ideas
      (pp. 211-224)

      During this period very many visitors came to the Moore School and left to spread the word about the new world of the computer. These people were of course all connected with the war effort but virtually all were so connected only for the duration. They had university positions and by this date were already conscious of the fact that the end of the war was in sight. They were the cadre around which the early computer people were formed.

      Perhaps the most important of the domestic visitors were various members of the Institute for Advanced Study, the NDRC’s Applied...

    • Chapter 10. First Calculations on the ENIAC
      (pp. 225-236)

      By the fall of 1945 the ENIAC was fast approaching completion. Gillon, who had been on an assignment in the Pacific, returned and Simon wrote Pender, saying: “No doubt you remember the stimulus for the new computing devices came largely from Gillon. The supervision of the contracts came to me because of Colonel Gillon’s departure. . . . Neither Dr. Dederick nor I really get the time . . . in fact I would really like to see the entire supervision of the contracts passed back to the Research Division of the Ordnance office so that Gillon would control it...

  8. PART THREE: Post-World War II:: The von Neumann Machine and the Institute for Advanced Study

    • Chapter 1. Post-EDVAC Days
      (pp. 239-251)

      As mentioned earlier, it was clear to me by the summer of 1945 that the development of computers had to continue and in a more normal peacetime mode. I therefore had a number of conversations with friends to gain some feeling for what was possible and desirable. Among those I talked to was John Kline, chairman of the mathematics department at the University of Pennsylvania and Secretary of the American Mathematical Society. Kline was both a wise and kindly gentleman, and he gave me much good advice.

      Of course, the most obvious choice for a location for this work was...

    • Chapter 2. The Institute for Advanced Study Computer
      (pp. 252-270)

      The staffing of the Institute project really began in March 1946 when I arrived full time in Princeton. The man who was for a number of years to be the chief engineer, Julian Bigelow, accepted the Institute’s offer on 7 March with the intention of coming full time to Princeton as soon as he could free himself from his duties. These consisted of responsibilities to the Fire Control Division of NDRC as well as its Applied Mathematics Panel. “He has a profound interest in automatic computing and control which is clearly a very important asset in this work. Part of...

    • Chapter 3. Automata Theory and Logic Machines
      (pp. 271-285)

      More or less simultaneously with his interests in logical design, planning and coding, and solving hydrodynamical problems for Los Alamos, von Neumann had a profound concern for automata. In particular, he always had a deep interest in Turing’s work and indeed offered Turing the position of his assistant at the Institute for Advanced Study in 1938.¹ Turing however decided to return to Kings College, Cambridge, where he was a Fellow.

      Before describing the developments with automata, we should perhaps say a little about logic machines, since they touch at least tangentially on our story. The reader who wants a good...

    • Chapter 4. Numerical Mathematics
      (pp. 286-299)

      The branch of mathematics concerned with numerical calculation is very old, going back at least to the day of Archimedes who had, along with so much else, a procedure for finding upper and lower bounds for square roots of both small and large numbers.¹ Many of the great names of mathematics, such as Newton and Euler, are attached to procedures for carrying out various numerical tasks. One of the greatest of these is the Prince of Mathematicians, Carl Friedrich Gauss (1777–1855). Bell says of him: “Archimedes, Newton, and Gauss, these three, are in a class by themselves among the...

    • Chapter 5. Numerical Meteorology
      (pp. 300-305)

      As we mentioned earlier, the aim of the Institute Electronic Computer Project was a threefold thrust into numerical mathematics, some important and large-scale application, and engineering. For the second effort von Neumann chose numerical meteorology. This choice was probably dictated by his profound understanding of hydrodynamics and by his desire to show the fundamental importance of the modern computer to our society. Moreover, he knew of the pioneering work during World War I in this field by Lewis F. Richardson, whom we shall discuss presently. Richardson failed largely because the Courant condition had not yet been discovered, and because high-speed...

    • Chapter 6. Engineering Activities and Achievements
      (pp. 306-320)

      The engineering activity at the Institute was the keystone in the arch built there. Without it, all the rest would have been pointless and in a very real sense incomplete. With it, the total contribution to society by the relevant Institute members has been of very large proportions.

      We were very fortunate indeed in having Julian H. Bigelow as our chief engineer during the formative stages of the project. He had visionary ideas on circuitry and the intellectual toughness to force these ideas to fruition. Without his leadership it is doubtful that the computer would have been a reality. All...

    • Chapter 7. The Computer and UNESCO
      (pp. 321-324)

      Of course while all this was going the rest of the world was not sitting quiescent. On the contrary, the computer revolution swept very rapidly throughout the United States, Europe, Israel, and Japan. Most of the work abroad was done at universities or research institutions and resulted in a large number of one-of-a-kind computers. None of these was in itself particularly important except that each served to condition some country to the need for electronic computers. Thus the integrated effect of all this activity was to create a demand for the electronic computer, first in the scientific and then a...

    • Chapter 8. The Early Industrial Scene
      (pp. 325-332)

      Now that we have at least in broad brush painted the worldwide computer scene in the academic and governmental communities as of 1957 or thereabouts, we must examine the industrial responses to the challenges of these communities. The academic and governmental activities both in the United States and throughout the world undoubtedly developed a substantial market-place for computers, not only for scientific purposes but also for commercial ones. None of these groups was however equipped to carry the field forward. Each made its contribution by its training of students, engineers, and scientists to appreciate and to use these new tools....

    • Chapter 9. Programming Languages
      (pp. 333-341)

      As I suggested earlier, one of the great contributions industry made to the computer field was the development of common lines of machines, so that users could potentially communicate with one another. Undoubtedly, this was a great impetus to the development of scientific languages such as FORTRAN. In the succeeding pages I shall discuss the programming developments of the period 1946-1957, since they foreshadow many of the modern ones.¹

      While the electronic computer produced a revolution by increasing incredibly the speed of processing data, it still left a large task for the human: the task of programming the problems to...

    • Chapter 10. Conclusions
      (pp. 342-348)

      Now that we have traced some of the ideas and met some of the people behind computer development in the period from 1623 through 1957, it is time we reflect a little on what we have learned in the way of underlying themes and trends.

      There are several such themes that suggest themselves at once: the fact that mankind chose to automate computing rather than some other phase of the human condition; that this area of human occupation should prove to be so extraordinarily productive and useful to man; that the automation of computation should alter irreversibly theWeltanschauungof...

  9. Appendix: World-wide Developments
    (pp. 349-362)
  10. Index
    (pp. 363-378)