George Stibitz

Progress is man’s ability to complicate simplicity.
Thor Heyerdahl

George Stibitz (1904-1995)
George Robert Stibitz (1904-1995)

On a late evening in November 1937, a research mathematician at the Bell Labs, George Stibitz, left his working place to go home, taking from the Bell stockroom two telephone relays, a couple of flashlight bulbs, a wire, and a dry cell. At home, Stibitz sat behind the kitchen table and started to assemble a simple logical device, which consisted of the above-mentioned parts and a switch, made from a tobacco tin. He soon had a device, which proved to be the first relays binary adder, in which a lighted bulb represented the binary digit “1” and an unlighted bulb, the binary digit “0.” His wife Dorothea named it the K-model, after “kitchen table”. The next day Stibitz took the K-model to the Labs to show the colleagues, and they speculated on the possibility of building a full-size calculator out of relays. His colleagues reasoned that any practical relay computer, using binary arithmetic, would need perhaps hundreds of relays, thus making it both bulkier and more expensive than the commercial mechanical calculators then in use at the Labs.

But what George Stibitz realized was, that a relay calculator could perform not just one but a sequence of calculations, with relay circuits directing the order and storing interim results as needed. Specifically, it could perform the sequence of operations required to perform multiplication and division of complex numbers: two mathematical operations that researchers elsewhere at the Bell Labs frequently performed in connection with filter and amplifier design for long-distance circuits. At Labs in the 1930s, a roomful of human “computers” figured complex number quotients and products using commercial mechanical calculators. The calculations themselves are straightforward enough: a complex multiplication requires about six simple arithmetic operations, while complex division requires about a dozen operations, and each requires temporary storage of a few intermediate results.

The K-model of Stibitz
The K-model of Stibitz

Stibitz did not know that in Berlin Konrad Zuse was doing almost the exact same thing at the same time. Stibitz however did know that Claude Shannon also had studied the correspondence of statements of symbolic logic with binary relay circuits while a graduate student at MIT. Shannon wrote his graduate thesis (published in 1938) on that subject and then went to Bell Labs, where he and Stibitz learned of each other’s work. But Shannon was not actively involved in the design of the computers of Stibitz. Clearly, the idea of using relays to implement binary logic was common in the late 1930s (a similar discovery was made in Japan).

When Stibitz first demonstrated his K-model computer to company executives, they were not very impressed. There were no fireworks, no champagne, as he remembered later on. Less than a year later, however, Bell executives had changed their minds about the Stibitz invention. An important factor in that decision was the increasing pressure on Bell to find a way of solving its increasingly complex mathematical problems. The company agreed to finance the construction of a large experimental model of Stibitz’s invention. Stibitz completed the designs in February 1938, and the construction of the machine began in April 1939, by Samuel Williams, a switching engineer in Bell. The final product was ready in October and was first put into operation on 8 January 1940, and remained in service until 1949. As Bell Labs built other relay computers during the war, its name was changed from the initial Complex Number Computer to Model 1. The cost was some 20000 USD.

Drawing of the teletype of the Complex Number Computer
Drawing of the teletype of the Complex Number Computer

Initially, the Complex Number Computer performed only complex multiplication and division, but later a simple modification enabled it to add and subtract as well. It used about 400-450 binary relays, 6-8 panels, and ten multiposition, multipole relays called “crossbars” for the temporary storage of numbers. The machine used the decimal system with the decimal point fixed at the beginning of each number. Internally, four binary relays coded each digit, using a code that represented a decimal digit n by the binary code for n+3; this simplified the problem of digit carry and subtraction (excess-three binary coded decimal is still called “Stibitz-code” today). The machine handled ten-digit numbers in its registers but displayed and printed eight-digit answers (range ±0.99999999). It used “prefix” notation: that is, operators keyed the arithmetic operations before they keyed in the operands. For example, to multiply the two complex numbers (2+3i) by (4+5i), the operator would key in (see the nearby drawing of the keyboard):
M +.2 +i .3 +.4 -i .5 =
The letter M stands for multiply (the letter D on the keyboard is for division). Note the location of the decimal point before each of the four numbers. The machine would actually be calculating (0.3+0.5i) x (0.4-0.2i), and printing the answer 0.07000000+i 0.22000000. The operator would have to scale the results accordingly (multiply by 100). A simple adding operation took about 100 mS, while the multiplication of two complex numbers took about forty-five seconds.

The calculating unit has four registers and is completely separated from the input/output unit, which is a special terminal (see the nearby photo). The computer itself was kept in an out-of-the-way room in the labs, where few ever saw it. The operators accessed it remotely using one of three modified teletype machines (keyboard and a printing device), connected to the processor by a multiple-wire bus and placed elsewhere, which however cannot work simultaneously.

Stibitz developed further the idea of remote, multiple access to a computer. On 11 September 1940, the American Mathematical Society met at Dartmouth College in Hanover, New Hampshire, a few hundred miles north of the building of Bell Labs in New York, where was the Complex Number Computer. Stibitz arranged to have the computer connected by telephone lines (28-wire teletype cable) to a teletype unit installed there. The Complex Number Computer worked well, and there is no doubt it impressed those who used it. The meeting was attended by many of America’s most prominent mathematicians, as well as individuals who later led important computing projects (e.g., John von Neumann, John Mauchly, Norbert Wiener, and Garrett Birkhoff). The Dartmouth demonstration foreshadowed the modern era of remote computing, but remote access of this type was not repeated for another ten years.

The calculating unit of the Model I
The calculating unit of Model I

The Complex Number Computer was not programmable. A combination of relay circuits permanently controlled its sequence of operations. Those relays were of the same type as the ones used to handle the numbers, but the machine did not have a separate, clearly defined part that handled the “control” of the computing sequence. (Later Bell Labs computers did.) The concept of programmability arose at Bell Labs only after the Complex Number Computer was built after its builders saw that its basic computing elements were unduly restricted by its marriage to control circuits tying it to nothing but complex arithmetic. (Besides complex arithmetic, they tried to get the machine to perform polynomial arithmetic, of which complex arithmetic is a special case. But the machine was too restricted for that.)

The success of the Complex Number Computer encouraged Stibitz to propose more ambitious designs that included the ability to modify the calculator’s operations by perforated tape. At first, the Labs turned down his proposals, but with the entry of the United States into the Second World War in December 1941, Bell Labs shifted its priorities toward military projects that involved more computation than its peacetime research. Most of their wartime accomplishments were in the design of analog computers. But they also built five digital relay computers for military purposes, and one more after the war’s end for their own use, making a total of seven digital machines counting the Complex Number Computer.

The first of these calculators for military use was the Relay Interpolator, installed in Washington, D.C. in 1943 and later known as the Model II. It was built from 440 relays and a memory capacity of 7 numbers. The speed of multiplication was 4 seconds (multiplication by repeated addition). It mainly solved problems related to directing antiaircraft fire, which it did by executing a sequence of arithmetic operations that interpolated function values supplied to the machine by paper tapes. Like the Complex Number Computer, it was a special-purpose machine; however, its arithmetic sequence was not Relay Calculators permanently wired but rather supplied by a “formula tape” (five-channel paper tape) cemented into a loop. Different tapes, therefore, allowed one to employ different methods of interpolation. Model II could not do much besides interpolation, but as interpolation is a process that lends itself to the solution of many problems in science and engineering the machine was kept busy by other government agencies long after the war ended.

The next two machines, designed by Stibitz—the Ballistic Computer and the Error Detector Mark 22 (later known as Models III and IV), were identical machines, the first installed in 1944 at Fort Bliss, Texas, and the second in early 1945 in Washington (each one cost 65000 USD). They contained some 1400 relays and had a memory capacity of 10 numbers. The speed of multiplication was 1 second (multiplication by table look-up). These machines also used paper tapes for data and formula input, with the arithmetic sequence supplied by a loop of paper tape. Models III and IV, like Model II, also solved problems relating to the aiming and tracking of antiaircraft guns. They were, however, more sophisticated machines, having the ability not only to perform interpolation but also to evaluate the ballistic equations describing the path of the target airplane and of the antiaircraft shell. An additional paper tape directed which of those functions the machine was to evaluate. Thus, Models III and N were the first of the Bell Labs digital calculators to have some degree of general programmability, although neither was a fully general-purpose calculator. Their memory and arithmetic units had modest capabilities: only six decimal digits of precision, a memory of ten numbers for each machine.

Relay equipment room of the Model V computer
Relay equipment room of the Model V computer

The largest computer in the series and the last, designed by Stibitz, was the Model V, of which Bell Labs built two copies for the military in 1946 and 1947. It was a huge (weight 10 tons) and very expensive (500000 USD) machine. Each contained over nine thousand relays and handled numbers expressed in scientific notation. The store could hold up to thirty numbers, and paper tape readers fed in both program steps and numerical data. Speed of multiplication 0.8 sec. The most interesting aspect of Model V’s design was that it had two separate arithmetic units, each capable of operating as an independent computer with its own memory registers and input-output devices. Small-scale problems could be run in pairs on the machine, saving time, while bigger problems could take over both processors. Associated with each processor (using the modern term) were fifteen memory registers, for a total of thirty for the whole machine. A master control unit directed instructions to one or both processors according to their availability. This control unit was separate from the control units in the processor that directed the sequence of arithmetic, memory, and input/output operations; it controlled the control, so to speak. (Stibitz called it a “superbranching” capability.) Thus in a very real sense, the Model V had what is now called an “operating system”-a control unit that supervises and manages the flow of work through a computer.

Besides programming power, the later Bell computers stressed extraordinary reliability. Relays, used as a basic element for logical and memory operations, have a tendency to fail intermittently. Should a piece of dust lodge itself between two relay contacts, that circuit will fail, though the rest of the relay will be fine. After a few cycles, the dust particle may shake itself loose, after which everything will return to normal. Thus an entire computation may be way off without any machine failure showing up during a diagnostic session.

Bell’s engineers designed computer circuits that checked themselves at every step of a computation. The circuits were designed not only to add, subtract, store numbers, and so on; they were also designed to check that they had done those things correctly and to stop the machine otherwise. Bell’s engineers were also guided by their experience in designing telephone circuits that had to operate long hours unattended in often hostile environments. Those circuits were designed to be repaired by semi-skilled technicians; telephone service would be terribly costly if an engineer had to be called in every time a phone line went down or a customer’s phone went dead. The Bell Labs Models II through VI used a system whereby not four but seven binary relays coded each decimal digit. They were divided into two groups of two and five relays; the decimal code was as follows:

Decimal digit
Relays
0 01 00001
1 01 00010
2 01 00100
3 01 01000
4 01 10000
5 10 00001
6 10 00010
7 10 00100
8 10 01000
9 10 10000

Bell Labs called this system a “bi-quinary” notation since the relays had a weight of either one or five. Actually, it is not a combination of those number bases; rather, it is a seven-bit, mixed decimal code. All the Bell Labs relay computers worked in decimal arithmetic. A special circuit was checked to see that two and only two relays were energized for each decimal digit. Another circuit checked that for each group one and only one relay was on—that prevented two separate errors from canceling each other out, although certain unusual combinations of malfunctions could go undetected.

Biography of George Stibitz

George Stibitz in the middle 1920s
George Stibitz in the middle 1920s

George Robert Stibitz was born on 20 April 1904, in York, Pennsylvania. He was the firstborn of Mildred Amelia (Murphy) Stibitz (1873–1967), a math teacher, and Rev. Dr. George Stibitz (1856-1944), a son of German emigrants, professor of theology, and pastor of Zion Reformed church. George had a brother—Earl E. (1914-1993), and two sisters—Mildred T. (1907-2000) and Eleanor (1918-2016).

George’s childhood was spent in Dayton, Ohio, where his father taught at a local college. Stibitz, an experimenter at heart, had been intrigued by electrical gadgets since childhood, an interest that on occasion must have dismayed his parents. As a boy of eight in Dayton, Ohio, he nearly set the house on fire by overloading the circuits with an electric motor given to him by his father.

Because of the interest in and aptitude for science and engineering that he had exhibited, George was enrolled at the experimental high school Moraine Park in Dayton, established by Charles Kettering, inventor of the first automobile ignition system.

For his undergraduate studies, Stibitz enrolled at Denison University in Granville, Ohio. After earning his bachelor of philosophy degree there in 1926, he went on to Union College in Schenectady, New York, where he was awarded his M.S. degree in 1927. After graduating from Union College, he worked lonesome as a technician at General Electric research labs in Schenectady for one year, before returning to Cornell University to begin his doctoral program. Stibitz received his Ph.D. in mathematical physics from Cornell in 1930.

Stibitz’s first job after graduation was as a research mathematician at the Bell Telephone Laboratories in New York City. His job there was to work on one of the fundamental problems with which modern telecommunication companies have to deal: How to carry out the endless number of mathematical calculations required to design and operate an increasingly complex system of telephones. At the time, virtually the only tool available to perform these calculations was the desktop mechanical calculator. It was obvious that this device would no longer be adequate for the growing demands of the nation’s expanding telephone network and the pioneering work of Stibitz on computers proved to be very important.

From 1941-1945 Stibitz served in the National Defense Committee, where he worked on important theoretical work dealing with computation. After the war, he decided not to go back to Bell Labs, but to start a scientific and academic career. From 1945 to 1954, Stibitz worked as a private consultant in Burlington VT, developing a precursor to the electronic digital minicomputer. He joined the Dartmouth faculty and applied computer systems development to a variety of topics in biomedicine in 1964. In 1966 Stibitz became a Full Professor, and in 1970 he became a Professor Emeritus.

Stibitz married on 1 September 1930, to Dorothea Lamson (1905-2007), the daughter of Dr. Charles Allen Lamson (1865-1930), an MD in New London, New Hampshire, with whom he had two daughters, Mary Gertrude and Martha Amelia.

George Robert Stibitz held 38 patents, excluding those assigned to Bell labs. His great contribution to Computer Science was his creation of the Complex Number Calculator, which first ran in January 1940. This was the world’s first example of remote job entry, a technique that revolutionized the dissemination of information through phones and computer networks. In 1965, Stibitz received the Harry Goode Award for lifetime achievement in engineering from AFIPS. Among the other awards he has received are the Harry Goode Award of the American Federation for Information Processing (1965), the Piore Award of the Institute of Electrical and Electronic Engineers (1977), and the Babbage Society Medal (1982). He was also the recipient of honorary degrees from Keene State College and Dartmouth College and was named to the Inventors Hall of Fame in 1983.

George Robert Stibitz died in his home in Hanover, New Hampshire, on 31 January 1995, at age of 90.