ENIAC
The First General-Purpose Electronic Computer and the Women Who Programmed It
- 15 chapters
- 24m
- Programming
- Free · no sign-up
The book explains how ENIAC worked, including its operation times and reliability issues. It covers the programming methods using patch cables, the role of six women programmers, and the machine's work on hydrogen bomb calculations. Chapters also detail ENIAC's influence on Monte Carlo methods and its connection to the EDVAC project.
This account reveals how ENIAC's development led to modern computing concepts. It traces the evolution from wartime necessity to stored-program computers. Readers will find this essential for understanding early computer history and the forgotten contributions of women in technology.
Listen
-
Read this chapter
In June 1941, the U.S. Army Ordnance Department relied on graduate students, Friden calculators, and a differential analyzer to calculate artillery firing tables. John Mauchly, not an expert in electronics, began exploring ways to accelerate these computations using electronic methods. He partnered with research associate J. Presper Eckert to design a new kind of computer. By August 1942, Mauchly had proposed a fully electronic machine intended to perform the calculations much faster. The project was funded by the United States Army Ordnance Corps, under the direction of Major General Gladeon M. Barnes, at a cost of approximately $487,000—equivalent to about $7 million in today’s money.
The U.S. Army Ordnance approved their plan, awarding the University of Pennsylvania a six-month research contract worth $61,700. Construction started in June 1943 under the code name “Project PX,” with John Grist Brainerd as principal investigator. Herman Goldstine secured funding and was assigned to oversee the project. Work began at the Moore School of Electrical Engineering in July 1943. By September, Eckert and Mauchly finished designing the computer. Assembly started in June 1944, and construction wrapped up in May 1945. Testing followed at the Moore School. In November 1945, Eckert, Mauchly, Brainerd, and Goldstine released the first confidential report on how the machine worked and how it was programmed.
The team supporting Mauchly and Eckert included Robert F. Shaw, Jeffrey Chuan Chu, Thomas Kite Sharpless, Frank Mural, Arthur Burks, Harry Huskey, and Jack Davis, each responsible for specific parts of the machine. Jean Jennings, Marlyn Wescoff, Ruth Lichterman, Betty Snyder, Frances Bilas, and Kay McNulty were the female mathematicians who did most of the programming. In 1946, the researchers left the University of Pennsylvania to form the Eckert–Mauchly Computer Corporation.
ENIAC was a large, modular computer made up of individual panels that handled different tasks. Twenty of these panels were accumulators, capable not just of adding and subtracting, but also storing a ten-digit decimal number. Numbers moved between these units through general-purpose buses, which were called trays back then. To reach its impressive speed, each panel had to send and receive numbers, do the math, store the result, and start the next step—all without any moving parts. The machine could also make decisions based on whether a result was positive or negative, which gave it great flexibility in how it processed information.
-
Read this chapter
By the time it shut down in 1955, ENIAC was a massive machine filled with 18,000 vacuum tubes, 7,200 crystal diodes, 6,000 relays, 70,000 resistors, 10,000 capacitors, and around 5,000,000 hand-soldered joints. It stood about ten feet tall, was three feet deep, and stretched a hundred feet long, weighing over thirty short tons. The machine occupied 300 square feet and used 150 kilowatts of power. Input came from an IBM card reader, and output went to an IBM card punch, which could send data to an IBM accounting machine like the IBM 405 for printed results. When it first launched, ENIAC had no built-in memory, but those punch cards served as external storage. In 1953, a 100-word magnetic-core memory from the Burroughs Corporation was added.
ENIAC used ten-position ring counters to store digits, with each digit needing 36 vacuum tubes, 10 of them dual triodes that made up the flip-flops in the ring counter. Arithmetic was done by counting pulses through these ring counters, and when a counter "wrapped around," it generated a carry pulse. This method was designed to electronically mimic how a mechanical adding machine worked, using digit wheels as its model.
ENIAC featured twenty accumulators, each capable of holding ten-digit signed numbers using ten's complement representation. These units could carry out five thousand addition or subtraction operations per second with any other accumulator or a constant transmitter. By connecting multiple accumulators at once, the computer could perform many operations in parallel, greatly increasing its potential speed.
ENIAC's design included a special multiplier unit that controlled four accumulators, enabling up to 385 multiplications per second. Another unit managed five accumulators for division and square root functions, handling as many as 40 divisions or three square roots each second. Though connecting the carry output from one accumulator to another allowed for double precision in calculations, the timing limitations within the carry circuits made it impossible to string together three or more units for even greater accuracy.
ENIAC had ten main units working together, with the initiating unit starting and stopping the machine. The cycling unit kept everything in sync. The master programmer handled loop sequencing, while the reader controlled an IBM punch-card reader. There was also a printer that managed an IBM card punch, and a constant transmitter. The remaining units were three function tables. Each played a role in how the computer processed information.
-
Read this chapter
The times for operations mentioned earlier are further detailed by references from Rojas and Hashagen, or alternatively from Wilkes, offering additional insight into how long certain tasks took on the machine. These accounts provide more precise information than what was previously stated, though they do differ somewhat in their specific timings. The variations highlight the complexity of recording and reporting early computer performance, especially when dealing with systems like ENIAC that pushed the limits of what was then possible. While the exact figures may not match across sources, each adds to our understanding of how quickly the first general-purpose electronic computer could process data.
ENIAC operated on a basic machine cycle lasting 200 microseconds, driven by 20 cycles of a 100 kHz clock in the cycling unit. This timing allowed for 5,000 cycles per second when processing 10-digit numbers. During each cycle, the computer was capable of writing a number into a register, reading a number from a register, or performing an addition or subtraction between two numbers.
Multiplying a 10-digit number by another 10-digit number took fourteen cycles, or 2,800 microseconds—about 357 operations per second. The process was slower if one of the numbers had fewer than ten digits. Each multiplication required a specific number of cycles depending on the size of the second number, with the total increasing by four cycles for each additional digit in that second number, up to ten.
Division and square roots on ENIAC took a specific number of cycles based on the digits in the result—13 times the number of digits plus 13, or 13(d+1). For example, if the quotient or square root had ten digits, it would take up to 143 cycles, which equated to about 28,600 microseconds, or roughly 35 operations per second. Wilkes noted in 1956 that a division with a 10-digit result required six milliseconds. If the result had fewer than ten digits, the process was quicker.
-
Read this chapter
ENIAC ran on radio tubes common to the era, including octal-base models, with decimal accumulators built from 6SN7 flip-flops. The logic functions relied on a mix of 6L7s, 6SJ7s, 6SA7s, and 6AC7s, while 6L6s and 6V6s handled the line driving, sending pulses through cables between the racks.
ENIAC functioned only about half the time due to several tubes burning out each day. High-reliability tubes didn’t become available until 1948. The majority of failures occurred during warm-up and cool-down cycles, when thermal stress was greatest on the tube heaters and cathodes. Engineers eventually brought the failure rate down to a more acceptable level, managing only one tube failure every two days. In a 1989 interview, Eckert recalled, “We had a tube fail about every two days and we could locate the problem within 15 minutes.”
-
Read this chapter
ENIAC was not a stored-program computer like today’s machines; instead, it was a massive collection of arithmetic units that required programs to be physically set up by hand. Programming involved plugging wires into a complex system and adjusting three portable function tables, each with 1,200 ten-way switches. The process of translating a problem into a working program took weeks, and even after the logic was worked out on paper, it could take days to load everything into ENIAC using cables and switches. Once running, programmers could step through the machine’s operations one by one to check for errors. A tutorial using an ENIAC simulator helps show how these early programs were structured.
The six women who programmed ENIAC—Kay McNulty, Betty Jennings, Betty Snyder, Marlyn Wescoff, Fran Bilas, and Ruth Lichterman—didn’t just figure out how to load programs into the machine; they came to understand how it actually worked. They could track down problems so precisely that they often identified which specific vacuum tube had failed, allowing technicians to replace it quickly.
-
Read this chapter
During World War II, the U.S. Army needed help computing ballistics trajectories, and many women were interviewed for the job. At least 200 were hired by the Moore School of Engineering to work as "computers." Six of them—Betty Holberton, Kay McNulty, Marlyn Wescoff, Ruth Lichterman, Betty Jean Jennings, and Fran Bilas—were chosen to program ENIAC. They worked on calculations for the Army's Ballistic Research Laboratory. Though men with the same education and experience were called "professionals," these women were labeled "subprofessionals," even though they held professional degrees in mathematics and were highly trained mathematicians.
These women were not just figures in photos beside the ENIAC machine, as some believed, but real programmers who worked on the first general-purpose electronic computer. Computer scientist Kathryn Kleiman discovered this through her own research, going against what a historian had told her. Some of the women never received recognition for their work during their lifetimes. After the war, they continued programming ENIAC, and their skills made them hard to replace. In the 1990s, Kleiman learned most weren’t invited to the 50th anniversary. She set out to find them and record their stories. “They were shocked to be discovered,” Kleiman says. “They were thrilled to be recognized, but had mixed impressions about how they felt about being ignored for so long.” She published a book in 2022 about the six women.
At the Moore School of Electrical Engineering, a group of about two hundred women worked as human computers, calculating complex formulas with mechanical devices for science and engineering projects. They didn’t just compute numbers—they learned how the machines functioned, studying logic, structure, and circuitry. This rare technical opportunity was especially significant for women in the mid-1940s. Among them, Betty Holberton, originally Snyder, went on to help create the first generative programming system known as SORT/MERGE. She also played a role in designing early commercial computers like the UNIVAC and the BINAC, working alongside Jean Jennings. Jean McNulty contributed by developing subroutines that expanded ENIAC’s capabilities.
Herman Goldstine chose the programmers, whom he called operators, from women who had already been calculating ballistics tables with mechanical calculators and a differential analyzer during ENIAC’s development. Under Goldstine and Adele Goldstine's guidance, these women examined the machine’s blueprints and physical layout to learn how to manage its switches and cables, because no programming languages existed yet. Though many considered programming merely clerical work and did not publicly acknowledge the programmers' role in ENIAC’s success, three of the original programmers—Jean Bartik, Kay McNulty, and Betty Snyder—have since been honored for their work. In 2020, three Army supercomputers were named Jean, Kay, and Betty, after those women respectively.
The job titles “programmer” and “operator” weren’t seen as real professions for women at first. During World War II, a labor shortage made it possible for women to enter the field, but they were not viewed as skilled workers. The idea was to use women to free up men for more advanced roles. The National Advisory Committee for Aeronautics said in 1942 that engineers benefited from having women handle calculations, since the women did the work faster and more accurately than the engineers themselves. The engineers admitted that their own experience was being wasted on repetitive tasks.
After the original six programmers, a team of a hundred scientists was brought in to keep working on ENIAC. Some of those scientists were women, including Gloria Ruth Gordon. Adele Goldstine wrote the first technical description of the machine.
-
Read this chapter
John von Neumann, who was working on the hydrogen bomb at Los Alamos National Laboratory, learned about ENIAC while it was still in its early stages. The project had been going for about a year when von Neumann became aware of it. In December 1945, ENIAC began calculating thermonuclear reactions using complex equations. That data helped support research into building a hydrogen bomb.
-
Read this chapter
ENIAC played a key role in making the Monte Carlo method widely used in science, especially in relation to nuclear weapons development. Scientists working on the original atomic bomb had relied on large teams of people—called "computers"—to do massive calculations about how far neutrons would travel through different materials. John von Neumann and Stanislaw Ulam recognized that ENIAC’s speed could greatly accelerate these computations. The success of this work demonstrated the power of Monte Carlo methods, showing their value in scientific research.
-
Read this chapter
On February 1, 1946, a press conference announced the completion of the ENIAC machine, and just two days later, on February 14, the public saw it demonstrated at the University of Pennsylvania. Elizabeth Snyder and Betty Jean Jennings worked on the trajectory program shown during the demo, though Herman and Adele Goldstine claimed credit for it. The machine was formally dedicated the next day, but the women who helped build and program it were not invited to the ceremony or the dinner that followed.
ENIAC was officially accepted by the U.S. Army Ordnance Corps in July 1946, after the original contract of $61,700 ballooned to nearly half a million dollars—about $9 million in today’s money. Just months later, on November 9, 1946, the machine was shut down for upgrades and moved to Aberdeen Proving Ground in Maryland. There, it was turned back on July 29, 1947, and ran continuously until October 2, 1955, when it was retired. It was replaced by more efficient machines like EDVAC and ORDVAC.
-
Read this chapter
Just a few months after ENIAC was revealed in the summer of 1946, the Pentagon gathered what they called “the top people in electronics and mathematics from the United States and Great Britain” for a series of lectures in Philadelphia. This effort was part of “an extraordinary attempt to jump-start research in the field.” The lectures were officially titled The Theory and Techniques for Design of Digital Computers, though they were more commonly known as the Moore School Lectures. Half of the talks were given by the creators of ENIAC.
ENIAC was a one-of-a-kind machine that was never built again, because its design was frozen in 1943 and lacked key innovations like storing programs. Eckert and Mauchly began work on a new computer called the EDVAC, which would be simpler and more powerful. In 1944, Eckert described a memory unit using mercury delay lines that could hold both data and instructions. John von Neumann, who was consulting at the Moore School, attended those meetings and later wrote up an incomplete set of notes titled First Draft of a Report on the EDVAC. These notes were shared by Herman Goldstine with government and educational groups, sparking interest in a new generation of machines, including EDSAC at Cambridge and SEAC at the U.S. Bureau of Standards.
-
Read this chapter
After 1947, improvements were made to ENIAC, including a basic stored-program system that used function tables as read-only memory. Programming was still done by setting switches, but the idea was developed further by Richard Clippinger and his team, as well as by the Goldstines, and it was part of the ENIAC patent. Clippinger discussed the instruction set with von Neumann. Clippinger had proposed a three-address architecture, but von Neumann suggested a simpler one-address design. In this setup, three digits from accumulator #6 served as the program counter, accumulator #15 was the main accumulator, and accumulator #8 functioned as an address pointer for reading data from the function tables. The rest of the accumulators—1 through 5, 7, 9 through 14, and 17 through 19—were used for data memory.
In March 1948, a new converter unit was installed on ENIAC, allowing programming through a reader using standard IBM cards. The following month, April, the first results were produced using these new coding methods to solve the Monte Carlo problem. After ENIAC was moved to Aberdeen, engineers built a register panel for memory, though it never functioned properly. A small master control unit was also added, which allowed the machine to be turned on and off.
In April 1948, the stored-program capability of ENIAC was demonstrated for the first time, running a program designed by Adele Goldstine for John von Neumann. The modification was developed by Betty Jennings, Clippinger, Adele Goldstine, and others. It reduced ENIAC’s speed by a factor of six and eliminated its ability to perform parallel computations. Still, it cut reprogramming time from days to hours, a change many deemed worthwhile. Analysis showed that because of the difference between electronic computation and electromechanical input/output speeds, most real-world problems were already I/O bound, even without using the machine’s original parallelism. Even after the speed reduction, computations would remain limited by input/output operations.
In early 1952, ENIAC got a new high-speed shifter that made shifting five times faster. Then in July 1953, the system gained a 100-word expansion core memory, using binary-coded decimal and excess-3 number representation. To handle this new memory, the machine was fitted with a new Function Table selector, a memory address selector, and pulse-shaping circuits. Three new orders were also added to the programming mechanism to support the changes.
-
Read this chapter
Computers as mechanical devices have existed for centuries, with evidence dating back to Archimedes’ time, including the Antikythera mechanism. But the modern computer era truly began in the 1930s and 1940s. That’s when machines started to take on the characteristics we recognize today as electronic computing. The developments of that period laid the foundation for what would become the first general-purpose electronic computers, changing everything about how humans process information.
ENIAC stood alongside machines like the IBM Harvard Mark I and the German Z3 as one of the first computers capable of running any sequence of math operations, though it didn’t read instructions from tape like those earlier machines did. Similar to the British Colossus, programmers configured ENIAC using plugboards and switches. What made ENIAC special was that it combined full programmability—meaning it could theoretically solve any computational problem—with electronic speed. The Atanasoff–Berry Computer, ENIAC, and Colossus all relied on thermionic valves, or vacuum tubes, for their operation. But where the Z3, ABC, and Colossus used binary arithmetic, ENIAC’s registers worked in decimal.
ENIAC, like the Colossus, couldn’t be reprogrammed without physical rewiring until April 1948. In June of that same year, the Manchester Baby became the first electronic stored-program computer when it ran its first program. The idea of storing both data and instructions in the same memory had been discussed during ENIAC’s development, but it wasn’t included at first because wartime needs demanded a quick completion. With only twenty storage locations, ENIAC’s memory was too small to hold both programs and data together.
-
Read this chapter
The Z3 and Colossus were developed independently during World War II, without knowledge of each other or of the ABC and ENIAC. Work on the ABC at Iowa State University ended in 1942 when John Atanasoff went to Washington, D.C., for Navy research, and the machine was later dismantled. The Z3 was destroyed by bombing raids in Berlin in 1943. The ten Colossus machines, part of the UK's secret war effort, stayed hidden until the late 1970s, though their capabilities were known among UK staff and selected Americans. ENIAC, however, was shown off to the press in 1946 and captured global attention. A public demo was created by Snyder and Jennings, who programmed it to calculate a missile's trajectory in 15 seconds—a job that would have taken weeks for human computers. All but two Colossus machines were dismantled in 1945; the last two were used by GCHQ to decrypt Soviet messages into the 1960s.
-
Read this chapter
In 1947, Mauchly and Eckert applied for a patent for ENIAC, which was granted in 1964 under U.S. patent number 3,120,606. But in 1973, a federal court case called Honeywell, Inc. v. Sperry Rand Corp. ruled the patent invalid. The court found that the inventors had based their work on John Atanasoff’s earlier design, the Atanasoff–Berry computer, or ABC, which was prototyped in 1939. The decision officially recognized Atanasoff as the inventor of the first electronic digital computer and placed that invention into the public domain.
-
Read this chapter
In 1996, to mark the 50th birthday of ENIAC, the University of Pennsylvania brought together a team to create a project called “ENIAC-on-a-Chip.” They built a tiny silicon computer chip that was 7.44 millimeters by 5.29 millimeters and had the same functions as ENIAC. Even though this new chip ran at 20 MHz—much faster than the original—it still lagged far behind the microprocessors of the late 1990s.
In 1997, the six women who did most of the programming of ENIAC were inducted into the Women in Technology International Hall of Fame. A 2010 documentary film titled Top Secret Rosies: The Female "Computers" of WWII by LeAnn Erickson explores their role. That same year, a short film called The Computers, directed by Kate McMahon, shared their story; it emerged from two decades of research by Kathryn Kleiman and her team through the ENIAC Programmers Project. Then in 2022, Grand Central Publishing released Proving Ground by Kathy Kleiman, a biography focusing on those six programmers and how they translated block diagrams and electronic schematics of the ENIAC—then still under construction—into programs meant to run on the machine once it was ready for use.
In 2011, to mark the 65th anniversary of ENIAC’s debut, the city of Philadelphia honored the moment by naming February 15 as ENIAC Day.
Read
Free to download, keep and share. For general information only — not professional medical, legal or financial advice. Please consult a qualified professional.