ENIAC World Wide Web
The first high-speed electronic automatic computer, the ENIAC, was put into operation at Aberdeen Proving Ground in 1947 and introduced an era of possibilities never before available to the field of computation. It was one of the most significant developments of our time because it enabled us to expand out knowledge in almost every field through limitless quantities of accurate, inexpensive computation performed at high speeds.
Prior to the advent of the ENIAC, however, were thousands of years of development to produce devices which could remove the drudgery of computation. The very first requirement was the invention of a means to record numbers. Various means were used in early times to do this. The Hebrews, Greeks, and Romans, for example, used the letters of their alphabets to represent numbers, but these systems did not lend themselves to easy computation of mathematical problems. Anyone familiar with the Roman numeral system can imagine the difficulty in multiplying CCXVII by XXIX to get MMMMMMCCXCIII.
In spite of clumsy systems of recording numbers Egyptians, Greeks and Romans were nonetheless able to perform computations. A method of finger computation evolved and at some time prior to 600 B.C. an ingenious calculating device, the abacus, was developed and used by them. This calculating device employed a place system and was built around the base ten. Strangely enough it seems never to have occurred to those people to write numerals in this same system. That idea was to come many years later from India.
The Roman-type abacus found its way to the Orient around the twelfth century A.D. and today it still remains the principal means of computing in China and Japan. It is a simple device and a skilled operator can add, subtract, multiply, and divide with amazing speed. In fact, the champion abacus operator of Japan some years ago defeated a man using an electric calculating machine. The achievement of such speed and accuracy on the abacus, however, requires considerable training, which is one disadvantage of the abacus.
The modern abacus consists of a rectangular frame with a number of rods or wires. These rods are divided into two unequal portions by a transverse bar. On the upper, smaller portion of each rod are two beads and on the lower portion five beads. (The modern Japanese abacus has one bead above and four beads below the bar.) The whole acts as a numerical register, each rod representing a decimal order as in our familiar "arabic" notation.
The chief disadvantage of the abacus is that the operator must perform his calculations mentally. The device merely stores his results step by step. It does not perform the actual calculations as does the modern desk calculator.
As mentioned previously, in spite of the use of the abacus by the Greeks and Romans the Western peoples did not develop a convenient system of numeration. This was accomplished by Hindu scholars nearly two thousand years ago. They developed the symbols and the place system which enables us to express any number, no matter how large or small it may be. This system is easy to learn and simple to use. It is built around the base ten, although any number might have been used. It is possible that the base of ten was selected because man had always used his fingers for counting.
The Hindu numerals with their place system were carried westward by merchants and scholars and finally reached Spain in the tenth century. However, it was not until the fifteenth century that they finally prevailed in the Western World and took their present form 0, 1, 2, 3, 4, 5, 6, 7, 8, and 9. This development opened the door to modern mathematics. Around 1600 John Napier, a Scottish mathematician, invented logarithms and also prepared a convenient multiplication table on pieces of wood or bone. "Napier's bones," as they were called, had no practical value but his logarithms did. Around 1632 William Oughtred in England made the first slide rule by inscribing logarithms on wood or ivory. The slide rule is an analog computer which is still widely used by scientists and engineers throughout the entire world.
Shortly after Napier's achievements, around 1642, Blaise Pascal in France constructed an adding machine that resembled the modern desk calculator. This machine consisted of a series of ratchet-driven wheels with the numbers 0 to 9 on them. Although it was a simple device and could only add and subtract it embodied one very important development which is an essential part of the digital computing technique, namely a means of making the "carry over" into an automatic process. It also indicated the possibility of multiplying numbers by successive addition. For the first time a machine was built that approached automatic computing because it removed a function from the operator to the machine.
In 1671 Gottfried Leibnitz, in Germany, invented the stepped wheel, a cylindrical drum having nine teeth of increasing length along its surface. When the drum was rotated, a gear sliding on an axis parallel to that of the drum engaged some of the teeth, thereby undergoing a corresponding number of rotation steps. The Leibnitz wheel made it possible to multiply quickly by repeated additions. From then onwards a variety of improved calculating machines were devised, all based on the original idea of wheels moved around in steps, and by the nineteenth century commercial models began to appear. The Thomas machine (France) embodied this principle and, starting in 1822, was the first to be manufactured for widespread use. It was a dependable, crank- operated, four-process machine. In 1885 Burroughs, in the United states, produced an adding machine and in 1902 the first machine with completely automatic multiplication and division was built by Rechnitzer, also in the United states. Several machines were designed in various countries and by 1920, with the fully automatic Monroe machine, we reach the stage of desk calculators familiar to us in banks, offices, and laboratories. These are the machines that do most of the world's computation, the modern ones being electrically driven and assuming more and more of the operator's functions.
These machines, however, are limited in scope and require the constant attention of a human operator to supply them with numbers and to record on paper the result of each stage of the calculation before proceeding to the next one. They do save time and mental effort but they are not fast enough where very large quantities of numerical data have to be handled. It must also be noted that they are not true forerunners of the modern electronic computers.
The principles of modern computers, albeit in mechanical form, were laid down by Charles Babbage in England. starting in 1812 he attempted to build an automatic mechanical calculator. He was familiar with Mullers idea for a difference engine (1786) and Jaques' idea for controlling a loom by means of punched holes (1775). He incorporated both ideas and actually made models of a difference engine which was card controlled. Technical problems of construction were too great for mechanical engineering of that period and the computer was never finished. Later, Babbage conceived an even more ambitious idea. He worked on the design of an analytical engine, which, had it been built, would have incorporated ideas familiar to the modern digital computer.
Babbage's new idea was to extend the capabilities of the difference engine so that it could not merely add and print, but also multiply, divide, and call for new data from its human operator. In order for the analytical engine to perform the tasks expected of it, it was necessary to express all the possible instructions to the machine in the form of stereotyped commands. Such a procedure is known as the logic or the logical design of a computer, and Babbage's engine was strikingly modern in regard to its logical design. Except for the need of a human attendant to read into the machine values from mathematical tables, this engine was logically parallel to most of the recent automatic computers. Naturally the speed of computation would have been far below that of the modern calculator (it could perform one addition per second), since Babbage had to use the purely mechanical techniques available in his day.
A significant part of his design for an analytical engine was the use of a "store" for holding the partial results of arithmetic operations performed in what he called the "mill." To feed numbers into the machine he proposed a system using patterns of holes punched in cards, while the results of calculations were to be presented by a mechanism which would set up the appropriate numerals in type ready for printing. Another proposal was that the engine should be capable of deciding on a course of action in the light of partial results obtained. He called this the "control." All these ideas have become features of the modern digital computer, and it is mainly because of them that our present-day machines are so fast in operation and so versatile in their range of calculations.
The next important development along this line was made by Hollerith and Powers during 1880 - 1890. Working for the United States Census Bureau they required some means of handling large masses of information. This led to the development of the first key punch, the first sorter, and the first tabulator, all using large punched cards. The patterns of round holes in the cards were formed by using a crude hand-punching device. The cards could be sorted into the series or order wanted, and then tabulated to reduce the census data to manageable form. Hollerith did not use the punched cards to control the course of computation, but simply used the holes in the cards to carry information. Hollerith and Powers also employed the electromagnetic techniques in their machine, that is, information on the card (hole or no hole) was converted into electric impulses, and the electric impulses actuated devices known as relays.
In 1925, at the Massachusetts Institute of Technology, Dr. Vannevar Bush and his associates began studying the problem of designing a machine to solve differential equations. In 1927, Dr. Bush started work on a mechanical differential analyzer. He experimented with mechanisms that would integrate, add, multiply, etc., and methods of connecting them together in a machine. Originally he made use of electrical currents flowing through watt-hour meters, which integrate the electrical power absorbed by circuits attached to them. Later he reverted to the mechanical integrators of the wheel and disc type. A major part of the success of the machine depended on a device whereby a very small turning force would do a rather large amount of work. By careful design of the discs and of the wheel mountings, Dr. Bush was able to make the integrators accurate, and, by introducing torque amplifiers to take the loads off the wheels so that they rotated very freely, he was able to reduce the slippage between the wheel and the disc. He combined a frame with the integrators, which contained movable shafts, changeable gears and couplings, input hand-wheels and output recording pens for producing a readable result.
By 1930 the first differential analyzer was finished. It was entirely mechanical, having no electrical parts except the motors. Its disadvantage was the great deal of work required in changing from one problem to another. This involved undoing old mechanical connections between shafts and setting up new ones. For this reason the design of a second analyzer was begun in 1935, on which all connections could be made electrically.
As early as 1932 personnel of the Ballistic Section at Aberdeen Proving Ground had investigated the possible use of the Differential Analyzer for ballistic computations. The machine was used to compute numerous trajectories including a set sufficient for the preparation of a complete range table. It was also tested on interior ballistic problems. A careful study and evaluation of the results of this investigation led to the following conclusions:
1. Such a machine could be built at Aberdeen Proving Ground.
2. The machine could perform various kinds of ballistic computations with the necessary accuracy.
3. Operation and maintenance would require no unusual skill.
4. The speed of the machine was such that it would insure greatly increased output of work by the Ballistic Section in any emergency.
5. It could perform routine work at a smaller total expenditure of funds.
6. Its use in the preparation of range tables would replace numerous complicated and indirect methods by simple, uniform and direct process.
7. Its use would permit the investigation of various problems, the solution of which was necessary to any advance in ballistic theory or practice but which previously could not be studied because existing methods required a prohibitive amount of computing labor.
It was decided to use the Bush Differential Analyzer at Aberdeen Proving Ground and its installation was completed in 1935. After final adjustment and trials it was used in the computation of two firing tables. These were completed with gratifying results by mid 1936.
The Bush Differential Analyzer, an analog device, was installed under the direction of Major James Guion, who was in charge of the Ballistic Computing Section at Aberdeen Proving Ground at that time. Major Guion had been quick to realize the value of this continuous variable calculator to the computational needs of the Ordnance Department. The availability of this machine and the experience that was gained in its use were invaluable to the Ordnance Department during the crucial transition period prior to America's entry into World War II.
The analyzer installed at APG had ten integrating units and provisions for two input and two output tables. However, despite its value as an important mechanical aid to computation it had certain limitations. The torque amplifier, for example, although simple in mechanical design, frequently failed toward the end of a long trajectory run with the loss of the preceding computation and an appreciable delay associated with its repair. This type of failure and the interference that it caused with the carrying out of computations at the Ballistic Research Laboratory led to extensive study to overcome this problem.
At the outbreak of World War II in Europe the officer in charge of ballistic computations at APG was Lt. P. N. Gillon. He immediately recognized the immensity of the task that would fall upon the Ordnance Department in the event of United States participation in the war and this prompted him to seek both marked improvement in mechanical aids to computation and augmented facilities for its accomplishment. A number of very important things were undertaken to accomplish this purpose.
The Moore School of Electrical Engineering of the University of Pennsylvania had a Bush Differential Analyzer of somewhat larger capacity than the one at Aberdeen Proving Ground. It had fourteen integrating units instead of ten. Therefore the Ordnance Department awarded a contract to the University of Pennsylvania for the utilization of this device. Several additional contracts were later awarded to the University to carry out different phases of the increasingly important role which it was to play in the computation activities of the Ordnance Department during the war.
Lt. P. N. Gillon, in his capacity as officer in charge of ballistic computations, conferred frequently with Dean Pender, Professor Brainerd, and their associates at the Moore School in order to effect proper coordination of the computational work at the two localities, Philadelphia and Aberdeen. There was a very talented group at the Moore School under the direction of Professor Brainerd and as a result of Lt. Gillon's discussions with them Assistant Professor Weygand undertook to develop an electronic torque amplifier to replace the mechanical torque amplifiers on the Bush Differential Analyzers. This work was eminently successful and, in addition, photoelectric followers were developed by the Moore School group for both the input and output tables of the analyzer. As a result of these accomplishments the productive capacity of the analyzers at both the Moore School and Aberdeen Proving Ground were enhanced by at least an order of magnitude.
During World War II the Bush Differential Analyzer was used primarily to compute trajectories for firing tables and to prepare trajectory charts for use with VT fuzes. The machine could compute a 60-second trajectory in about 15 minutes, whereas a human operator using a desk calculator required about 20 hours to perform the same computation. Its value to the Ordnance Department during the war years was tremendous.
The next major development, prior to the advent of the high-speed electronic automatic computer, came in 1944 when Howard Aiken, in cooperation with some IBM engineers and some graduate students at Harvard University, completed the Mark I Relay Computer, also called the Automatic Sequence- Controlled Calculator.
This machine performed thousands of calculating steps, one after another, according to a scheme fixed ahead of time. Its individual operations were automatic, once the punched tape fixing the chain of operations had been put on the machine, and it was sequence-controlled, since control over the sequence of its operations was built into the machine.
The principle of operation of the Mark I Relay Computer was electro- mechanical. It combined the two applications of the concept of state, that is, the information for processing was represented by patterns of open and closed relays (like the patterns of holes and no holes which Babbage used).
Although the machine was efficient, fast, and capable of solving a variety of problems its speed could not approach that of the electronic computer, which is the next state in the development of computers and the subject of this monograph. Nevertheless, it holds a permanent place in the history of development of calculating machines, since it was the first automatic machine to be completed.
(BRL at APG used two types of relay computers during 1947-1954. Two IBM Relay Calculators were used for a short time but were not successful. Two Bell Relay Computers were used. They were accurate, but slow and required expert maintenance. Dust and humidity adversely affected their operation. However, many useful results were obtained from the twin Bell Relay Computers during many years of successful operation. One good feature of the Bell Relay Machines was that they would run 16 hours overnight, unattended. A problem regularly would be placed on the computer at 4:00 PM. The next morning, arriving operating personnel would find the problem solved and the machine waiting, or the machine still feverishly running the problem to a successful conclusion. Many times, if it was felt that the machines would complete a problem during the night, a new problem was set before the machine for it to tackle until personnel arrived the next morning. This mode of operation increased the productivity of the computers with a minimum of personnel. Of course, once in a while a dirty relay would cause a stoppage, for the Bell A and the Bell B would either run without error or not run at all.)