Atanasoff’s thirty-five-page paper was intended to do what merely watching the ABC work could not do—to demonstrate to the Iowa State College Research Corporation that the computer was innovative, powerful, and successful. The goal was more money—Atanasoff and Berry estimated that they needed $5,000 to go on to the next step. The original and three carbon copies were made of the paper. Atanasoff sent one to the research corporation, one was retained by Berry for his use in overseeing the construction of the next prototype, and the third was set aside for the patenting process that Atanasoff thought the machine was ready for.
In December 1940, Atanasoff took his family east for a vacation, which would include his attending the annual meeting of the American Association for the Advancement of Science, held in Philadelphia. According to Burton, Atanasoff’s intense concentration on his teaching, on the ABC, and on his defense project had taken a toll on his marriage, but Joanne Atanasoff, the second daughter, and John Vincent II did feel comfortable riding their bikes to the physics building to see their father—they often played around in the basement there while he worked.
The automobile trip to the East Coast would be a long one, with stops in New York City and Washington, D.C., as well as Philadelphia. Atanasoff planned to do some work—mostly patent research in New York and Washington. Berry intended to meet them. John and his family celebrated Christmas in a hotel in New York, just a little dazzled by the urban world they were not accustomed to.
1. According to Alice R. Burks and Arthur W. Burks, “The model had two storage bands, each with twenty-five condensers, on the outerfaces of a large disk. One brand represented the abacus … or the counter drum, the other that of the keyboard drum. It had an add-subtract mechanism, served by a single carry-borrow condenser … and also a mechanism to perform the restore but not the shift function” (p. 22).
2. Yet another irony of Atanasoff’s story is that the student who dismantled the computer, Robert Stewart, later served as chairman of the computer science department.
Chapter Five
Throughout the Second World War, the Germans used a mechanical encoding device that they called the Enigma machine. It had been patented in 1918 or 1919 and put to use by the German army and navy by 1929. In 1931, a German working in the Cipher Office began selling information about the machine (including photographs of the instruction manuals) to the French, but neither the French nor British could break the code. It was a Pole, Marian Rejewski, who first cracked the German Enigma code in 1932 and built a replica of the machine. The Poles were then able to decode Wehrmacht radio messages until the late thirties—advances in Enigma technology foiled them as of 1937 for naval messages, and as of December 1938 for the rest of the German messages. The decoding machines Rejewski constructed were called “Bombas” (named, some said, after the ticking sounds they produced while working). The Bombas operated according to Rejewski’s insight that German intelligence operators signaled the day’s encryption key by typing in the same three letters twice in a row (for example NGHNGH) followed by the new settings for the three rotors of the Enigma machine. Knowing what these double letters signified, Rejewski then inferred the entire structure of the Enigma and its operation—the Bombas were built to sift through strings of code and find those that were likely to be messages. Through mid-1939, the Poles kept their knowledge to themselves. When the Germans introduced more rotors into the Enigma, the Poles quickly figured out how the rotors worked, but five rather than three rotors raised the number of possible combinations tenfold, outstripping the capacity of the Bombas to quickly sort through encoded messages. At the same time, the political and military situation in Poland was rapidly deteriorating, so the Poles communicated what they had discovered about the decoding of the Enigma to English and French intelligence. Rejewski and his fellow cryptographers spent the war sometimes in France, sometimes in Gibraltar, and sometimes in England, working with Allied intelligence.
Cracking the Enigma code was especially crucial for the British, since it was the code used by the German navy, and Britain was dependent on ocean traffic for every kind of supply, and therefore especially vulnerable to naval disruption or blockade. When Turing first arrived shortly after the invasion of Poland and the British declaration of war, six Bombas at Bletchley Park sifted through intercepted messages for matching letters that would reveal the settings of the German positions encoding the messages, and most of the code breaking was done by linguists, not mathematicians. The prized form of cryptanalytic intelligence was the sort that solves puzzles through a combination of linguistic sophistication and intuition. Turing was an enthusiastic puzzle solver, but since he was also a mathematician, he understood both large numbers (as in the number of combinations of letters that had to be tested in order to break a code) and probability (which combinations were likely to lead to dead ends and which were likely to be productive). It was Turing and an associate, Gordon Welchman, who were to address the problem of the extra rotors that had been added to the Enigma machine. The new “Bombes,” as they were rechristened, were designed using relays. Andrew Hodges maintains that Turing “was the right person to see what was needed, for his unusual experience with the relay multiplier [he had built at Princeton] had given him insight into the problems of embodying logical manipulations in this kind of machinery.” For his part, Welchman redesigned the wiring that constituted the instructions for the machines.
Andrew Roberts points out in The Storm of War that code breaking was not the only form of intelligence that the Allies were using even at the beginning of the war—more traditional methods such as spying, interrogating, and eavesdropping were also employed, but to break the codes meant they could listen to exchanges of information and instruction in real time, and so throughout the war, the code breakers were considered, in Churchill’s words, “the geese who laid the golden eggs” and “never cackled.”
When Turing went to Bletchley Park in September 1939, Germany seemed to have all the advantages: Stalin had signed a nonaggression pact on August 23, and the Russian army invaded Poland from the east two weeks after Germany invaded from the west. At the end of November, the Russians invaded Finland. Just after the declaration of war, the Germans had attacked an English ocean liner, the SS Athenia, killing 112 or 117 passengers (depending on the source). With the declaration of war, U-boats began steadily harassing English ships—on September 17, an aircraft carrier, the HMS Courageous, was sunk by two U-boats and went down in fifteen minutes, losing five hundred men. Historian Andrew Roberts notes that “by the end of 1939, Britain had lost 422,000 tons of shipping” by means of attacks and mines and was in danger of being isolated, without resources or even food if the German navy could manage it. The first half of 1940 was worse in every way: Finland fell, Norway fell, the first because of the passivity of the Allies, the second in spite of the Allies’ efforts. In May, the Dutch surrendered and English troops were driven back to Dunkirk, only to be evacuated, according to Roberts, because Hitler overruled the wishes of his generals, Kleist and Guderian, with a “halt order” that prevented them from pursuing and wiping out the retreating armies. France fell at the end of June, and the Battle of Britain began in July. Since the United States had declared its neutrality, the situation for Britain was desperate.
Turing’s Bombe (the first went to work in March 1940) was constructed like a large, heavy bookcase, six and a half feet high, more than seven feet long, and two and a half feet deep. The “books” were rows of motorized rotating drums, ends facing outward, twenty-six letter positions inscribed around the circumference of each. These were meant to simulate the operations of Enigma rotors. The Bombe worked as a sorter, trying out likely combinations of letters supplied by the operator to see if any German words were created as the patterns of letter correlations were changed. Most of the time, according to Turing’s mathematically based insight, sets of letters supplied by the operators (known as “cribs”) would proceed by logical substitutions to a state of self-contradiction. If the opera
tor would suspect that A = K, for example, when it arrived at a position in which A = A, it would be self-replicating and not correct, since the decoders knew that for Enigma, a letter could not be encoded as itself. As the rotors tested the positions, they could throw out any self-replicating positions they arrived at. The greatest number of positions that had to be tested for any letter was twenty-five, the fewest, one. Each Bombe contained stacks of rotors that tested the letter combinations simultaneously. The Enigma in Germany was operated by hand, but the Bombe was motorized, so that even though Enigma encoding positions were changed every night at midnight, the Bombes (eventually there were 211) could sort through probable encoding patterns very quickly. When combinations that looked fruitful were found, the code wheels on the English replica of the Enigma machine were set to mimic what had been found, and either a message came up or it didn’t. The code breaking was painstaking and tedious work that was aided by captures of German equipment or mistakes on the part of German personnel, as well as the tendency of the German military to use set phrases and clichéd expressions. In December 1939, Turing was instrumental in deciphering five days’ worth of five rotor codes from 1938, thus demonstrating that the codes could be broken and showing how. In January 1940, Turing was dispatched to France to meet with Rejewski and his colleagues. On January 17, Turing and the Poles succeeded in breaking codes from the previous October. Throughout 1940, the code breakers made progress, aided by the capture of code wheels from a U-boat in February and another set in November. Fortunately for the British, though the Germans knew that the U-boats had been destroyed, they did not realize that the code wheels had been salvaged.
Once the Bombes had proved successful, Turing had another idea—a machine that would take the output of the Bombe and bypass the work of human decipherers by automatically translating that output from code into understandable German. It was in order to implement this idea that Tommy Flowers came to Bletchley from the General Post Office. But Flowers and Turing never succeeded in putting together that particular machine, in large part because as the war progressed, it turned out that Enigma was not British intelligence’s biggest challenge.
Konrad Zuse, now an enlisted man in the German army, was still pursuing his own interests. Early on, Kurt Pannke wrote his commander a letter, asking that Zuse be relieved of duty because the invention he was working on would be valuable to the war effort, especially the Luftwaffe. This letter succeeded only in offending Zuse’s commanding officer, who did not believe that the Luftwaffe needed any help. Zuse used the army as a place to take up chess and think about his computer theory and offered himself to work on coding and decoding, but the Germans considered that that problem had been solved by Enigma and the other machines they had devised (see chapter 6). Then Zuse’s friend, Schreyer, attempted to get authorization to work on the computer for air defense, but when he suggested that research and development might take two years, the official in charge exclaimed, “What do you mean, after we’ve already won the war!” Finally, Zuse was put to work as a structural engineer working on weapons at Special Division F, with Henschel Aircraft. The task was to develop remote-controlled bombs. One type was to be dropped from an airplane and controlled by radio until it reached its destination. Another was to be dropped into water, where it would act as a torpedo. Toward the end of the war, the division worked on defensive surface-to-air missiles.
Zuse continued to develop his computer in the evenings and on weekends, managing to bring the second version of his computer, the Z2, to the demonstration stage in 1940—though it wasn’t always reliable. Zuse reports in his autobiography that only hours before the planned demonstration took place, the Z2 could not be made to work, but then, once the audience for the demonstration had arrived, it “performed flawlessly,” only to become temperamental again once the demonstration was over. He remarks that “afterwards, I hardly ever got the Z2 to run smoothly.” The problem, he felt, was not necessarily the design, per se, but that all available relays were secondhand parts from different manufacturers that had to be reconfigured to work in the way Zuse wanted them to, and that in reworking them, he overlooked details of how they would function together. But the single flawless demonstration aroused the interest of the technical director of the Aeronautics Research Institute, which was enough to gain Zuse a contract to develop the Z3. The contract meant money, but the war effort meant that he still had to use secondhand parts.
The Z3, which was completed in 1941, did work reliably. It incorporated the following principles and design ideas:
Electromagnetic relay technology (not vacuum tubes)
Binary number system
Floating point (a system of locating the decimal point)
Word length: 22 bits
Storage capacity: sixty-four words
Control by means of eight-track punched tape
Input by means of specially designed keyboard
Output by means of display of results on a row of lights, including proper placement of the decimal point
High speed: 3 seconds for multiplication, division, or square root
John Gustafson, who constructed the replica of the ABC and is an expert on early computers, writes:
It was a jaw-dropping accomplishment to invent floating-point arithmetic back then and get it to work at such high speed. It wasn’t just a way of adjusting the decimal point: he could represent positive and negative infinity, undefined numbers like 0/0, and a number of other ideas that did not become standardized until the 1980s. Not many computer engineers today, given a pile of electromagnetic relays, would have the faintest idea how to build a floating-point unit out of it, especially not one that can take square roots. He was very far ahead of his time. It is also worth noting that the 64 words of memory were addressable; the computer could pick out a particular one to use by its number. The ABC didn’t have anything like that—the ABC had memory in the two drums, but the operator selected which one to use, while on the Zuse machine, it was controllable by the program tape. It was like a modern computer in almost every way except that it couldn’t do conditional branches, which is testing a number and then jumping to a different part of the program depending on whether the test was true or false.
Gustafson adds, “This is why I admire Zuse every bit as much as I admire Atanasoff … and why I’m thankful that the arrogance of the German military didn’t see the merit of Zuse’s work, since the world might be a very different place now if they had.”
Zuse continued to make do with what he could find—since he could not get hold of a tape-punching machine, he punched strips of celluloid film with a manual hole punch; since his relay coils were secondhand, they were not uniform, so he had to adjust the voltage of each one in order to get them to work together.
Even though Zuse was making progress, and he could demonstrate the usefulness of his machine for certain calculations having to do with wing flutter in airplanes, he could not prove that his computer work was valuable to the war effort. He was put back into the army again in 1941 but managed to establish that his work with Henschel was worth a deferment. His work on the computer progressed, still on his own time.
Wartime rules and regulations favored Zuse’s machine in some ways. After persuading Henschel to let him work part-time, he set up his own company to develop the computer. He writes:
Available were unskilled, mostly female workers, who had made themselves unpopular elsewhere, or who did not fit into the normal working world. So, at one time I was able to hire an excellent technical designer who had had a lengthy stay in a mental hospital. In a normal company, his eccentricities probably would have gotten on everyone’s nerves, but we didn’t have any problems with him … My book-keeper had done something very foolish when he was a young man, and he had been prosecuted for it. But he fit in perfectly with our small company … There was also the great added advantage that neither of these workers could be drafted.
One of Atanasoff’s goals in attending the meeting of the American Ass
ociation for the Advancement of Science in December 1940 was to find out what other inventors were doing—he still feared that some larger, more prestigious, and better financed entity might be onto ideas similar to his. He was not giving a presentation himself, though.
One scholar, a man named John Mauchly, gave a talk about correlating weather patterns with solar phenomena such as sunspots, a subject that Atanasoff was interested in (as he was interested in allergies, soybeans, goat milk products, and home construction). In the course of his lecture, Mauchly, who was the only physics professor at Ursinus College in Collegeville, Pennsylvania, mentioned that he had devised a calculator, which he called the “Harmonic Analyzer,” to do the correlations. He detailed his design ideas and talked about his plans for building a more powerful machine. Although the Harmonic Analyzer was an analog machine, Mauchly said in his talk that he thought the future of computing was electronic, and he expected to have an electronic machine in about two years.
John Mauchly was about four years younger than Atanasoff. His background was middle-class academic, more like that of John Hasbrouck Van Vleck than like that of Stibitz, Aiken, or Atanasoff. Mauchly’s father, Sebastian, was a principal at a high school in Cincinnati, Ohio, until 1916, when John was nine and Sebastian received his PhD in physics and moved to Chevy Chase, Maryland, to become chief physicist specializing in “electricity and earth currents” at the Carnegie Institute in Washington, D.C. One thing he was interested in was the physics of lightning strikes (presaging John’s interest in weather prediction). John Mauchly was something of a prodigy and a pest, like the young Atanasoff. According to Scott McCartney in ENIAC, he had a sign over his bed that read, “What should I be doing now?” In 1919, Chevy Chase was a fairly new suburb, home to many men employed in scientific fields around the Washington, D.C., area. Twelve-year-old John, who as a five-year-old had rigged a flashlight out of a battery and a lightbulb, laid intercom wires in the trenches that workmen were digging for water lines. He was also a night owl who concocted a switch that turned off his reading light if one of his parents stepped onto the landing outside his door. In high school, he was an impressive student who planned to follow in his father’s footsteps as a physicist. By the time John was ready to go to college in 1925, though, Sebastian Mauchly had come to understand, possibly from his own experience, that there was more money in engineering than in physics, so John applied for and received a prestigious scholarship to Johns Hopkins University in engineering. But he got bored with that after about two years and transferred to the physics department, where he so impressed his professors that they decided to put him directly into the PhD program. He completed his doctorate in 1932, writing his dissertation on carbon monoxide. Here, his experience, and the conclusions he drew from it, also mirrored Atanasoff’s experience two years earlier—the calculations, which he performed on a Marchant desk calculator, proved onerous and inspired the ambition to invent a more powerful calculator.