The Z1 did, however, show that the logical concept Zuse had designed would work in theory. A college friend who was helping him, Helmut Schreyer, urged that they make a version using electronic vacuum tubes rather than mechanical switches. Had they done so right away, they would have gone down in history as the first inventors of a working modern computer: binary, electronic, and programmable. But Zuse, as well as the experts he consulted at the technical school, balked at the expense of building a device with close to two thousand vacuum tubes.²⁶

So for the Z2 they decided instead to use electromechanical relay switches, acquired secondhand from the phone company, which were tougher and cheaper, although a lot slower. The result was a computer that used relays for the arithmetic unit. However, the memory unit was mechanical, using movable pins in a metal sheet.

In 1939 Zuse began work on a third model, the Z3, that used electromechanical relays both for the arithmetic unit and for the memory and control units. When it was completed in 1941, it became the first fully working all-purpose, programmable digital computer. Even though it did not have a way to directly handle conditional jumps and branching in the programs, it could theoretically perform as a universal Turing machine. Its major difference from later computers was that it used clunky electromagnetic relays rather than electronic components such as vacuum tubes or transistors.

Zuse’s friend Schreyer went on to write a doctoral thesis, “The Tube Relay and the Techniques of Its Switching,” that advocated using vacuum tubes for a powerful and fast computer. But when he and Zuse proposed it to the German Army in 1942, the commanders said they were confident that they would win the war before the two years it would take to build such a machine.²⁷ They were more interested in making weapons than computers. As a result, Zuse was pulled away from his computer work and sent back to engineering airplanes. In 1943 his computers and designs were destroyed in the Allied bombing of Berlin.

Zuse and Stibitz, working independently, had both come up with employing relay switches to make circuits that could handle binary computations. How did they develop this idea at the same time when war kept their two teams isolated? The answer is partly that advances in technology and theory made the moment ripe. Along with many other innovators, Zuse and Stibitz were familiar with the use of relays in phone circuits, and it made sense to tie that to binary operations of math and logic. Likewise, Shannon, who was also very familiar with phone circuits, made the related theoretical leap that electronic circuits would be able to perform the logical tasks of Boolean algebra. The idea that digital circuits would be the key to computing was quickly becoming clear to researchers almost everywhere, even in isolated places like central Iowa.

JOHN VINCENT ATANASOFF

Far from both Zuse and Stibitz, another inventor was also experimenting with digital circuits in 1937. Toiling in a basement in Iowa, he would make the next historic innovation: building a calculating device that, at least in part, used vacuum tubes. In some ways his machine was less advanced than the others. It wasn’t programmable and multipurpose; instead of being totally electronic, he included some slow mechanical moving elements; and even though he built a model that was able to work in theory, he couldn’t actually get the thing reliably operational. Nevertheless, John Vincent Atanasoff, known to his wife and friends as Vincent, deserves the distinction of being the pioneer who conceived the first partly electronic digital computer, and he did so after he was struck by inspiration during a long impetuous drive one night in December 1937.²⁸

Atanasoff was born in 1903, the eldest of seven children of a Bulgarian immigrant and a woman descended from one of New England’s oldest families. His father worked as an engineer in a New Jersey electric plant run by Thomas Edison, then moved the family to a town in rural Florida south of Tampa. At nine, Vincent helped his father wire their Florida house for electricity, and his father gave him a Dietzgen slide rule. “That slide rule was my meat,” he recalled.²⁹ At an early age, he dove into the study of logarithms with an enthusiasm that seems a bit wacky even as he recounted it in earnest tones: “Can you imagine how a boy of nine, with baseball on his mind, could be transformed by this knowledge? Baseball was reduced to near zero as a stern study was made of logarithms.” Over the summer, he calculated the logarithm of 5 to the base e, then, with his mother’s help (she had once been a math teacher), he learned calculus while still in middle school. His father took him to the phosphate plant where he was an electrical engineer, showing him how the generators worked. Diffident, creative, and brilliant, young Vincent finished high school in two years, getting all A’s in his double load of classes.

At the University of Florida he studied electrical engineering and displayed a practical inclination, spending time in the university’s machine shop and foundry. He also remained fascinated by math and as a freshman studied a proof involving binary arithmetic. Creative and self-confident, he graduated with the highest grade point average of his time. He accepted a fellowship to pursue master’s work in math and physics at Iowa State and, even though he later was admitted to Harvard, stuck with his decision to head up to the corn belt town of Ames.

Atanasoff went on to pursue a doctorate in physics at the University of Wisconsin, where he had the same experience as the other computer pioneers, beginning with Babbage. His work, which was on how helium can be polarized by an electric field, involved tedious calculations. As he struggled to solve the math using a desktop adding machine, he dreamed of ways to invent a calculator that could do more of the work. After returning to Iowa State in 1930 as an assistant professor, he decided that his degrees in electrical engineering, math, and physics had equipped him for the task.

There was a consequence to his decision not to stay at Wisconsin or to go to Harvard or a similar large research university. At Iowa State, where no one else was working on ways to build new calculators, Atanasoff was on his own. He could come up with fresh ideas, but he did not have around him people to serve as sounding boards or to help him overcome theoretical or engineering challenges. Unlike most innovators of the digital age, he was a lone inventor, drawing his inspiration during solo car trips and in discussions with one graduate student assistant. In the end, that would prove to be a drawback.

Atanasoff initially considered building an analog device; his love of slide rules led him to try to devise a supersize version using long strips of film. But he realized that the film would have to be hundreds of yards long in order to solve linear algebraic equations accurately enough to suit his needs. He also built a contraption that could shape a mound of paraffin so that it could calculate a partial differential equation. The limitations of these analog devices caused him to focus instead on creating a digital version.

The first problem he tackled was how to store numbers in a machine. He used the term memory to describe this feature: “At the time, I had only a cursory knowledge of the work of Babbage and so did not know he called the same concept ‘store.’ . . . I like his word, and perhaps if I had known, I would have adopted it; I like ‘memory,’ too, with its analogy to the brain.”³⁰

Atanasoff went through a list of possible memory devices: mechanical pins, electromagnetic relays, a small piece of magnetic material that could be polarized by an electric charge, vacuum tubes, and a small electrical condenser. The fastest would be vacuum tubes, but they were expensive. So he opted instead to use what he called condensers—what we now call capacitors—which are small and inexpensive components that can store, at least briefly, an electrical charge. It was an understandable decision, but it meant that the machine would be sluggish and clunky. Even if the adding and subtracting could be done at electronic speeds, the process of taking numbers in and out of the memory unit would slow things down to the speed of the rotating drum.