Other articles in the series:
- The History of Relays
- The History of Electronic Computers
- The History of the Transistor
- The History of the Internet
The second project for the creation of an electronic computer, which emerged as a result of the war, like the 'Colossus,' required numerous minds and hands for its fruitful realization. But, just like the 'Colossus,' it would never have come into existence without one single person being obsessed with electronics. In this case, his name was .
Mauchly's story intertwines in mysterious and suspicious ways with that of John Atanasoff. As you may recall, we left Atanasoff and his assistant Claude Berry in 1942. They had abandoned their work on the electronic computer to pursue other military projects. Mauchly had much in common with Atanasoff: they were both physics professors at lesser-known institutions that lacked prestige and authority in broad academic circles. Mauchly languished in isolation as a teacher at tiny Ursinus College in the suburbs of Philadelphia, which had nowhere near the modest prestige of Iowa, where Atanasoff worked. Neither of them did anything to attract the attention of their more elite peers, say, from the University of Chicago. However, both were captivated by an eccentric idea: to build a computing machine from electronic components, the same parts used in radio and telephone amplifiers.
John Mauchly
Weather Prediction
For a while, these two men established a certain connection. They met in the late 1940s at an American Association for the Advancement of Science (AAAS) conference in Philadelphia. There, Mauchly presented his research on cyclical patterns in weather data using an electronic harmonic analyzer he had developed himself. This was an analog computer (meaning it represented values not in digital form but as physical quantities, in this case, current — the more current, the greater the value), functioning similarly to the mechanical tide predictor designed by William Thomson (later to become Lord Kelvin) in the 1870s.
Atanasov, sitting in the hall, knew he had found a companion for his lonely journey into the realm of electronic computations, and without hesitation, he approached Mauchly after his lecture to tell him about the machine he built in Ames. But to understand how Mauchly ended up on stage with his presentation of the electronic weather computer, one must return to his roots.
Mauchly was born in 1907 to physicist Sebastian Mauchly. Like many of his contemporaries, as a boy he became interested in radio and electronic tubes, and he wavered between a career as an electronics engineer and a physicist before deciding to focus on meteorology at Johns Hopkins University. Unfortunately, after graduating, he landed directly in the clutches of the Great Depression, and he was grateful to get a job at Ursinus in 1934 as the sole member of the physics faculty.
Ursinus College in 1930
At Ursinus, he embarked on his dream project — deciphering the hidden cycles of the global natural machine and learning to predict the weather not for days, but for months and years ahead. He was convinced that the Sun governed weather patterns lasting for several years, linked to solar activity and sunspots. He wanted to extract these patterns from the vast amount of data accumulated by the American Meteorological Bureau using students and a set of desktop calculators bought for pennies from bankrupt banks.
It soon became clear that there was too much data. The machines could not perform calculations quickly enough, and human errors began to appear when constantly copying the intermediate results from the machines onto paper. Mauchly started thinking about another approach. He knew about counters based on vacuum tubes, first created by Charles Wynn-Williams, which his fellow physicists used to count subatomic particles. Given that electronic devices could clearly record and accumulate numbers, Mauchly became intrigued by the idea of them performing more complex calculations. For several years, he played with electronic components in his spare time: switches, counters, substitution cipher machines that used a mix of electronic and mechanical components, and a harmonic analyzer that he applied to a weather prediction project, extracting data similar to multi-week patterns of rainfall fluctuations. This discovery led Mauchly to the AAAS in 1940, and later, Atanasoff to Mauchly.
Visit
The key event in the relationship between Mauchly and Atanasoff occurred six months later, in early summer 1941. In Philadelphia, Atanasoff told Mauchly about the electronic computer he had built in Iowa, mentioning how cheaply it had cost him. In their subsequent correspondence, he continued to make intriguing hints about how he built his computer for no more than $2 per digit. Mauchly was intrigued and quite surprised by such an achievement. By that time, he was already harboring serious plans to build an electronic calculator, but without support from the college, he would have to pay for all the equipment out of his own pocket. One tube typically cost $4, and storing one binary digit required at least two tubes. How, he wondered, had Atanasoff managed to save so much?
After six months, he finally had time to travel west to satisfy his curiosity. After one and a half thousand kilometers by car, in June 1941, Mauchly and his son visited Atanasoff in Ames. Mauchly later recounted that he left feeling disappointed. Atanasoff's cheap data storage was not electronic at all but relied on electrostatic charges on a mechanical drum. Because of this, and other mechanical components, as we have seen, he could not perform calculations at speeds even approaching those that Mauchly dreamed of. Later, he referred to it as a 'mechanical trinket that used a few electronic tubes.' However, shortly after the visit, he wrote a letter praising Atanasoff's machine, stating that it was 'electronic in nature, and could solve any system of linear equations with no more than thirty variables in just a few minutes.' He claimed it could be faster and cheaper than mechanical. Bush.
Thirty years later, the relationship between Mauchly and Atanasoff would become crucial in the Honeywell versus Sperry Rand lawsuit, resulting in the annulment of patents for the electronic computer developed by Mauchly. Without discussing the merits of the patent itself, although Atanasoff was a more experienced engineer and considering Mauchly's suspicious opinion of Atanasoff's computer expressed in hindsight, there is no reason to suspect that Mauchly learned or copied anything significant from Atanasoff's work. More importantly, the ENIAC design has nothing to do with the Atanasoff-Berry computer. The most that can be claimed is that Atanasoff bolstered Mauchly's confidence by demonstrating that an electronic computer could work.
Moor and Aberdeen School
At that time, Mauchly found himself back where he started. There was no magic trick for affordable electronic storage, and as long as he stayed in Ursinus, he lacked the resources to bring his electronic dream to life. Then fortune smiled upon him. That same summer in 1941, he attended a summer course in electronics at the Moore School of Engineering at the University of Pennsylvania. By then, France was already occupied, Britain was under siege, submarines roamed the Atlantic, and America's relations with aggressive expansionist Japan were quickly deteriorating [and Nazi Germany attacked the USSR / note by the translator]. Despite the isolationist sentiments among the populace, American intervention seemed possible and, likely, inevitable to elite groups from places like the University of Pennsylvania. The Moore School offered a course for engineers and scientists to accelerate their training for potential military work, particularly in radar technology (radar shared features with electronic computing: it used vacuum tubes to create and count high-frequency pulses and the intervals between them; however, Mauchly later denied that radar had a significant impact on the development of ENIAC).
Moore School of Engineering
The course led to two major outcomes for Mauchly: first, it connected him with John Presper Eckert, known as Pres, from a local real estate magnate family, and the young wizard of electronics who spent all his days in the lab of television pioneer . Later, Eckert would share the patent (which was subsequently declared invalid) for ENIAC with Mauchly. Secondly, it secured Mauchly a position at the Moore School, ending his long academic isolation in the quagmire of Ursinus College. This seemingly happened not due to any particular merits of Mauchly, but simply because the school desperately needed people to replace the scholars who had gone on to work on military contracts.
By 1942, the majority of the Moore School was working on a military project: calculating ballistic trajectories using a combination of mechanical and manual work. This project naturally grew out of the existing connection between the school and the Aberdeen Proving Ground, located 130 km further down the coast in Maryland.
The range was established during World War I to test artillery, replacing a previous range in Sandy Hook, New Jersey. In addition to live fire tests, its role included calculating firing tables used by artillery in combat. Air resistance made it impossible to determine the landing point of a projectile simply by solving a quadratic equation. Nonetheless, high accuracy was crucial for artillery fire, as initial shots tended to cause the greatest damage to enemy forces—after which the enemy quickly took cover.
To achieve such precision, modern armies compiled detailed tables that informed gunners how far their projectile would land after being fired at a specific angle. Compilers used the initial velocity and position of the projectile to calculate its location and speed over a short time interval, then repeated those calculations for the next interval, and so on, hundreds and thousands of times. For each combination of gun and projectile, such calculations needed to be conducted for all possible firing angles, taking into account various atmospheric conditions. The computational workload was so extensive that in Aberdeen, all calculations for the tables, initiated after the end of World War I, were only completed by 1936.
Clearly, Aberdeen needed a better solution. In 1933, it contracted with the Moore School: the army would fund the construction of two differential analyzers, analog computers designed based on a scheme from MIT led by . One would be sent to Aberdeen, while the other would remain at the Moore School for use at the discretion of the faculty. The analyzer could construct a trajectory in fifteen minutes, a calculation that would take a person several days, though the accuracy of the computer's calculations was slightly lower.
Demonstration of the howitzer at Aberdeen, circa 1942
However, in 1940, the research division, now called the Ballistic Research Laboratory (BRL), requested its machine that had been at Moore School and began calculating artillery tables for the impending war. The school's computing group was also enlisted to support the machine with human calculators. By 1942, 100 women calculators at the school worked six days a week, grinding out calculations for the war — among them was Mauchly's wife, Mary, who worked on the Aberdeen firing tables. Mauchly was placed in charge of another group of calculators working on calculations for radar antennas.
From the day he arrived at Moore School, Mauchly promoted his idea of an electronic computer throughout the faculty. He already had significant support from Presper Eckert and , a senior faculty member. Mauchly provided the idea, Eckert the engineering approach, Brainerd the persuasiveness and legitimacy. In the spring of 1943, this trio decided it was time to pitch Mauchly's long-nurtured idea to Army officials. But the enigmas of the climate that he had long sought to unravel would have to wait. The new computer was to serve the needs of a new master: to track not the eternal sine waves of global temperature cycles, but the ballistic trajectories of artillery shells.
ENIAC
In April 1943, Mauchly, Eckert, and Brainerd drafted the 'Report on the Electronic Differential Analyzer'. This brought another ally into their ranks, , a mathematician and Army officer who served as a liaison between Aberdeen and Moore School. With Goldstein's assistance, the group presented the idea to the committee at BRL and received a military grant, with Brainerd as the project's scientific director. They needed to complete the machine by September 1944 with a budget of $150,000. The team named the project ENIAC: Electronic Numerical Integrator, Analyzer, and Computer.
From left to right: Julian Bigelow, Herman Goldstein, Robert Oppenheimer, John von Neumann. The photo was taken at the Institute for Advanced Study in Princeton after the war, with a later model of the computer.
Like the Colossus in Britain, the authoritative engineering leadership in the United States, such as the National Defense Research Committee (NDRC), was skeptical about the ENIAC project. Moore School did not have the reputation of an elite educational institution, but it proposed to create something unprecedented. Even industrial giants like RCA struggled to develop relatively simple electronic counting circuits, let alone a configurable electronic computer. George Stibitz, the architect of relay computers at Bell Labs, who was then working on the NDRC project, believed that it would take too long to create ENIAC for it to be useful in the war.
In this, he was right. Creating ENIAC would take twice as long and three times more resources than initially planned. It drained a significant portion of the human resources of the Moore School. Just for the development alone, an additional seven people were needed beyond the initial group of Mauchly, Eckert, and Brainerd. Like the Colossus, ENIAC attracted many human calculators to help set up their electronic replacement. Among them were also Herman Goldstine's wife, Adele, and Gene Jennings (later Bartik), who would play a significant role in computer development. The letters NI in the name ENIAC suggested that the Moore School was providing the army with a digital, electronic version of the differential analyzer, which would solve integrals for trajectories faster and more accurately than its analog mechanical predecessor. But the result turned out to be something much greater.
Some ideas for the project may have been borrowed from a proposal made by Irwin Travers in 1940. It was Travers who participated in signing the agreement for the Moore School to use the analyzer in 1933, and in 1940 he proposed an improved version of the analyzer, though it was not electronic, it operated on a digital principle. It was supposed to use mechanical counters instead of analog wheels. By 1943, he had left the Moore School and took a position in fleet management in Washington.
The foundation of ENIAC's capabilities, much like the 'Colossus', lay in its variety of functional modules. Most often, accumulators were used for addition and counting. Their design was borrowed from the electronic counters of Wynn-Williams, utilized by physicists, essentially performing addition akin to how preschoolers count on their fingers. Other functional modules included multipliers, function generators, which sought data in tables, thereby replacing the counting of more complex functions like sine and cosine. Each module had its own programming settings that defined a small sequence of operations. Like the 'Colossus', programming was carried out using a combination of a panel with switches and switch panels resembling telephone exchanges.
ENIAC had several electromechanical components, particularly a relay register that served as a buffer between electronic accumulators and IBM punch card machines used for input and output. This architecture closely resembled that of the 'Colossus'. Sam Williams from Bell Laboratories, who collaborated with George Stibitz on the creation of Bell relay computers, also built a register for ENIAC.
A key difference from the 'Colossus' made ENIAC a more versatile machine: the ability to program its main settings. The main programmable unit sent pulses to the functional modules, triggering pre-set sequences, and received response pulses upon completion of tasks. It would then move to the next operation in the main control sequence and output the required calculations as a function of multiple smaller sequences. The main programmable unit could make decisions using a stepper motor: a ring counter that determined which of the six output lines to redirect the pulse to. This way, the device could execute up to six different functional sequences depending on the current state of the stepper motor. This flexibility allowed ENIAC to tackle problems far removed from its original ballistic focus.
Setting up ENIAC using switches and relays
Eckert was responsible for ensuring that all the electronics in this monster buzzed and hummed, and he invented the same basic tricks that Flowers did at Bletchley: the tubes had to operate on currents much lower than nominal, and the machine didn’t need to be turned off. However, due to the huge number of tubes used, one more trick was necessary: the plug-in modules, on each of which several dozen tubes were mounted, could be easily removed and replaced in case of failure. The servicing personnel would then leisurely find and replace the burnt-out tube, and ENIAC would be ready to work again immediately. Even with all these precautions, given the vast number of tubes in ENIAC, it could not perform calculations for an entire weekend or overnight as relay computers did. At some point, a tube would inevitably burn out.
Example of multiple tubes in ENIAC
Reviews of ENIAC often mention its enormous size. Rows of racks filled with tubes — there were a total of 18,000 — switches, and relays would occupy a typical country house and its lawn. Its size was due not only to its components (the tubes were relatively large), but also to its peculiar architecture. And although all mid-century computers seem large by modern standards, the next generation of electronic computers was much smaller than ENIAC and had greater capabilities using only one-tenth of the electronic components.
Panorama of ENIAC at the Moore School
The grotesque size of ENIAC stemmed from two main design decisions. The first sought to increase potential speed at the cost of complexity and expense. After this, nearly all computers stored numbers in registers and processed them in separate arithmetic modules, saving results back in the registers. ENIAC did not separate storage and processing modules. Each number storage module was also a processing module capable of addition and subtraction, which required many more vacuum tubes. It could be viewed as a highly accelerated version of the human calculators at the Moore School, as "its computational architecture resembled twenty human calculators working with ten-digit desktop calculators, passing calculation results back and forth." In theory, this allowed ENIAC to perform parallel calculations across multiple accumulators, but this capability was used sparingly, and by 1948 it was entirely eliminated.
The second design decision is harder to justify. Unlike ABC or Bell's relay machines, ENIAC did not store numbers in binary form. It translated decimal mechanical calculations directly into electronic form, with ten triggers for each digit — if the first was lit, that was zero, the second was 1, the third was 2, and so on. This was a huge expenditure of costly electronic components (for example, to represent the number 1000 in binary requires 10 triggers, one for each binary digit (1111101000); whereas in ENIAC's scheme it required 40 triggers, ten for each decimal digit), which apparently was organized out of fear of potential complexities in converting between binary and decimal systems. However, the Atanasoff-Berry computer, the 'Colossus,' and Bell's and Zuse's relay machines used the binary system, and their developers had no difficulties with conversions between bases.
Such design solutions will not be replicated by anyone. In this sense, ENIAC was similar to ABC — a unique oddity rather than a template for all modern computers. However, its advantage lay in the fact that it proved, beyond any doubt, the functionality of electronic computers by performing useful work and solving real tasks with astonishing speed for those around it.
Rehabilitation
By November 1945, ENIAC was fully operational. It could not boast the same reliability as its electromechanical counterparts, but it was reliable enough to leverage its speed advantage of several hundred times. A ballistic trajectory calculation that took a differential analyzer fifteen minutes could be completed by ENIAC in twenty seconds — faster than the projectile itself flies. And unlike the analyzer, it could do this with the same accuracy as a human calculator using a mechanical calculator.
However, as Stibitz predicted, ENIAC arrived too late to aid in the war, and the calculation of tables was no longer urgently required. But in Los Alamos, New Mexico, a secret weapon development project was ongoing, which continued even after the war. A lot of calculations were needed there as well. One of the physicists from the Manhattan Project, Edward Teller, had been inspired by the idea of 'superweapons' back in 1942: far more destructive than what was later dropped on Japan, with the explosion energy coming from nuclear fusion rather than fission. Teller believed he could initiate a fusion chain reaction in a mixture of deuterium (regular hydrogen with an extra neutron) and tritium (regular hydrogen with two extra neutrons). But this required a low tritium content, as it was extremely rare.
Therefore, a scientist from Los Alamos brought to Mura School calculations for verifying the superweapon, which required calculating differential equations modeling the ignition of a deuterium and tritium mixture for various tritium concentrations. No one at Mura School was authorized to know the purpose of these calculations, but they dutifully entered all the data and equations provided by the scientist. The details of the calculations remain classified to this day (as does the entire program for building the superweapon, now better known as the hydrogen bomb), although we know that Teller considered the results obtained in February 1946 as confirmation of the viability of his idea.
That same month, Mura School unveiled ENIAC to the public. During the opening ceremony in front of important figures and the press, the operators pretended to turn on the machine (although it was of course always on), conducted a few ceremonial calculations to compute a ballistic trajectory, demonstrating the unprecedented speed of the electronic components. After that, the staff distributed punched cards containing those calculations to everyone present.
ENIAC continued to solve several more practical problems throughout 1946: a set of fluid flow calculations (for example, for the airflow around an aircraft wing) for British physicist Douglas Hartree, another set of calculations for modeling nuclear weapon implosion, and trajectory calculations for a new ninety-millimeter gun in Aberdeen. Then it went silent for a year and a half. At the end of 1946, under a contract between Mura School and the army, BRL packed the machine and transported it to the testing range. There, it consistently suffered from reliability issues, and the BRL team could not get it to work well enough to perform any useful work until a major upgrade was completed in March 1948. We will discuss the upgrade, which completely revamped ENIAC, more in the next part.
But that no longer mattered. No one cared about ENIAC. The race to create its successor was already underway.
What else to read:
• Paul Ceruzzi, Reckoners (1983)
• Thomas Haigh, et. al., Eniac in Action (2016)
• David Ritchie, The Computer Pioneers (1986)
Source: habr.com
