War Machines: Women’s Computing Work and the Underpinnings of the Data-Driven State, 1930

Hidden within the story of Bletchley is a less popular narrative that cannot leverage the appeal of a lone genius and his accomplishments. Thousands of women worked at Bletchley during the war—most in technical roles.³ Although it is generally accepted that the striking and wide-ranging roles of the mostly women workers within Bletchley Park give lie to stereotypes about computing as a traditionally masculine field, the contributions of these women have not been analyzed as constitutive of larger trends in the history of computing. Instead, these workers have been positioned as a “reserve” labor force impelled into masculine-coded jobs due to the exigencies of war—and dispensed with immediately thereafter.⁴ Looking at the labor of women computer operators during World War II as though they were a temporary and completely exceptional type of workforce, however, hides important historical connections. US-centric narratives of computing and of the war also make it difficult to take the labor performed by these British women seriously; with the US experience as a backdrop, these women simply look like high-tech Rosie the Riveters. When taken on their own merits in the context of Britain’s experience of the war and the decades immediately before and after, however, this narrative can reorient our conceptions about computing’s development in Britain and internationally.

Women’s integration into the earliest computing systems both during and after the war shows their labor as formative to the project of computing and the twentieth-century technological state in a way that remains understudied. As our current understanding of computing technologies has grown to focus on how, why, and who deploys the power of cutting-edge technologies, the history of labor has become integral to the history of computing. With this attention to labor has come a renewed focus on the gendered dimensions of work with technology—a focus that resonates deeply with current concerns about underrepresentation in science, technology, and engineering. The history of women’s role in computing helps explain gendered categories that still construct labor forces in the Anglo-American context today.⁵ Yet this history is often perceived as a specialized, parallel narrative, rather than a foundational one. When women do not fulfill the role of inventor or entrepreneur in a way comparable to the men who have up until now been the main focus of computing history, their labor is often regarded as not being integral to the main narrative of computing’s history.⁶

This chapter analyzes how feminized work—work that was assumed to be rote, deskilled, and best suited to women—was critical in defining early computerization, from the inception of modern, pre-electronic computational methods in the early twentieth century, through the crucible of high-speed wartime codebreaking, to the postwar transition in government and industry that civilianized computing systems and their technological goals. This purportedly feminized work was anything but deskilled, and the needs, power, and money of the British government were instrumental in using it to create modern computing and computer workers before, during, and after World War II. Computing’s early beginnings as a feminized field presaged specific gendered labor hierarchies in peacetime—ones that put computing work at the bottom of the white-collar labor pyramid until the rise of technocratic ideals in the 1960s that reshaped the expectations and status of machine workers.⁷ Far from merely being artifacts of wartime pressures that appeared suddenly and disappeared with the coming of peace, these labor patterns defined British computing and its possibilities throughout the twentieth century, gendering the edifice of the technological state.

Creating Information Workers

Unlike the heavy manufacturing and munitions work into which women were drafted during World War II, computing held prewar associations with women’s labor. As in other industries, feminized “reserves” of labor shaped the viability and success of computing endeavors.⁸ Both British and American projects employed women from the outset as human computers—that is, workers who performed manual calculations—and then, very often, as computer operators and programmers. In Britain, women often built the machines as well. The manufacturing workforce at IBM UK was so feminized that management measured its production in “girl hours” rather than “man hours” into the 1960s.⁹

Prior to World War II, most women who worked as human computers did their work using a variety of desktop calculating machines. Working as clerks or scientific assistants, they made tables for civilian and military applications or did accounting tasks. At the same time, more complex machines and systems of calculation were being developed. Trained as an astronomer, the New Zealand–born Leslie John Comrie employed women workers to produce astronomical tables with the aid of general-purpose electromechanical punched card machines.¹⁰ While working at the British Nautical Almanac Office, Comrie earned a reputation as the father of large-scale application of punched card machines to scientific computing. Comrie’s major insight was that with effective organization general-purpose office machines could be used for complex scientific calculations—essentially creating an electromechanical supercomputer system from standardized punched-card components and well-trained labor.

Soon after, Comrie deployed his system for military applications, doing contract table-making and statistical work for the British government during World War II. Young women workers were key to his success, and he did not shy away from detailing why they were an ideal fit for his system. In his “Careers for Girls,” Comrie notes that women have the capacity to be easily trained to perform this work, along with the secretarial or typing work they might already do in the same office.¹¹ In addition, their numbers in the labor force were plentiful and, Comrie asserted, they diligently did work that young men saw as boring, dead-end drudgery. Like many other men in charge of computing systems, Comrie saw women’s labor not as an add-on but as foundational to his systems’ success.

During the early twentieth century, managers had gradually come to see young women as ideal office machine operators and had begun employing and training them to run a variety of machines. The growing popularity of the typewriter in late Victorian offices had helped create an association of automation with feminized labor pools. An emblematic document, prepared by Comrie to help government and business users compare the features of Hollerith and Powers-Samas punched card accounting machinery, described the operators of accounting machines as “girls from fifteen to eighteen or twenty years of age, who receive perhaps thirty or forty schillings a week for their work, and in some cases much less” (italics mine).¹² By determining labor costs and hiring patterns, gendered ideals influenced equipment purchasing and created the structure and expectations of modern office work. These women could have short, serviceable careers “before they (or most of them) graduate to married life and become experts with the housekeeping accounts!” Comrie quipped.¹³ Turnover through marriage was supposed to ensure women didn’t tire of the work or require promotion to better—and better paid—work, as that would throw his system out of alignment.

Figure 1.1 The “Powers Girl” was an advertising mascot of Powers-Samas who portrayed women’s labor as an integral part of mechanized accounting systems. *Vickers News*, January 1951.

Yet women workers often did not agree to the limited career paths set out for them, and sometimes young women in these roles pushed back against expectations. In the British Civil Service, Comrie’s ideas defined many women’s working lives, but these workers did not merely accept their lot. A flustered Treasury official reported in 1941: “Mrs. Arrow and I visited the Admiralty today. They have 180 [machine] operators, some established, but the majority temporary, and these girls are kicking up something of a fuss.”¹⁴ His report detailed how low pay and poor promotion prospects were creating an untenable situation and how managers had few outlets to appease the disgruntled workers given how the work was organized. Since the work had been deskilled, there was little room to advance.

Systems like Comrie’s did more than just slot people into socially appropriate work roles. They disciplined workers in accordance with certain gendered and classed labor ideals predicated on the heteronormative concept of a male breadwinner wage and unpaid domestic work for women within the nuclear family. These ideals were often not as fitting or practical as they seemed, resulting in more and more young women who had to support themselves and their families on near-poverty wages, particularly as the gendered contours of office labor continued to shift. In 1881, under 3 percent of all clerks were women. By 1921, this figure had rocketed to over 46 percent. By 1940, and continuing after the war, women held a clear and growing majority.¹⁵ In both the private and public sector, women clustered in low-level office jobs. Often, this work was referred to as subclerical—that is, below real clerical work, an idea later echoed by the informal appellation pink-collar labor. Defined by a lack of authority over other workers as much as by an assumption of skill-lessness, subclerical work increasingly became seen as ideal for young women in the twentieth century. It was the ultimate holding pattern for temporary labor, and only during times of crisis were women likely to break out of it.

Subclerical work utilized machinery heavily. Machines represented the incursion of manual, vaguely industrial work into the supposedly intellectual sphere of the office. As a result, clerks who performed this work suffered from a perception that what they did was not really integral to the real work of offices. Yet offices of the early to mid-twentieth century relied on machine workers in order to meet immediate demands and to define long-term goals. This style of machine-aided work quickly became essential to maintaining and expanding the powerful bureaucracies that constructed both government and industry. Many large institutions could no longer simply add to their number of workers to meet new objectives. Without mechanized systems and the cheap, plentiful labor pool of women who ran them, the reach and structure of these institutions could not scale.

Gendered War Work and High-Stakes Electronic Computing

As firsts in computing go, the Colossus computers stand out as being the first successful application of high-speed computing to accomplish time-sensitive, mission-critical goals. While the ENIAC was still under construction on the other side of the Atlantic, the second Colossus was ensuring the success of the Allies in the D-Day landings. Lone codebreakers, notably Alan Turing, receive much of the credit for British intelligence operations during World War II, but the most important and voluminous intelligence work of the war was machine-aided and feminized. Codebreaking operations employed thousands of women and hundreds of machines in addition to the elite cadre of men who led the work.

Along with the revolutionary codebreaking computers built by London Post Office researchers and operated by the Women’s Royal Naval Service (WRNS), work at Bletchley Park—the nerve center of the British intelligence establishment—also relied heavily on systems of punched card machines and women’s labor, like those designed by Comrie before the war. When commissioning Hollerith equipment for the park, Ronald Whelan reported that “following the setting up of the initial equipment in Hut 7 the workload increased rapidly.” In order to meet labor demands for punch and machine operators, “recruitment was made of local girls” who were trained in-house and worked the machines in shifts around the clock. Later in the war, “a great increase in operating strength came when a large component of WRNS personnel was drafted to us, in addition to a contingent of 20 university graduates.”¹⁶ Workers like these, many of whom were university educated or trained to a high standard at the park, would become the sinew in the information arm of the British war machine.

In times of crisis, the association of women with machine work strengthened and extended, opening up opportunities. Although wartime work as a whole upset social and economic hierarchies, the prewar association of women with electromechanical machine work contributed powerfully to wartime labor organization and, later, to the postwar labor landscape. In this context, working with the first digital, electronic, programmable computers the world had seen was not an unlikely role for many of the women workers who volunteered or were conscripted to work for the war effort.

In late 1941—in the same year that the Admiralty machine operators were “kicking up a fuss” about their pay and promotion prospects—the government began conscripting women into war work, registering and then calling up all British women between the ages of eighteen and sixty. Both married and single women were drafted into compulsory service across the home front. Assignments were made irrespective of locale: “You must be prepared to leave home unless this would mean exceptional hardship,” warned government announcements.¹⁷ Beginning in 1942, all young women were required to seek jobs only through the government’s Women’s Employment Exchange, set up to channel prime workers into essential industries.¹⁸

These women sometimes worked jobs that would have been unavailable to them in peacetime, and they also traveled and lived in ways that ran counter to peacetime social and familial mores. Nonetheless, the expectations and ideals of peacetime society—particularly in regard to a patriarchal, heteronormative family—continued to influence how and where women were deployed and who was called up. Single or childless widowed women between the ages of twenty and thirty were regarded as the most portable and dependable, followed by married women without children. Married women who had children under the age of fourteen escaped conscription or industrial reassignment; their service within the home was seen as more important. They were encouraged to volunteer, however.

Most women who worked for the war effort worked unusual hours or swing shifts, and either commuted far from home alone or relocated to spartan dormitories set up by the Ministry of Labour. Many of the women working at the heart of Britain’s intelligence establishment at Bletchley Park were housed in the servants’ quarters of a nearby country manor, with four bunks filling each small room beyond their intended capacity.¹⁹ Others streamed in by train each day. Women’s labor formed the nucleus of the British wartime information machine. They staffed the majority of posts in the information processing hub at Bletchley and at its related information collection outposts.

By 1942, all job training centers had been filled with women, and government documents noted that the shortage of manpower required a total reliance on women trainees from then on.²⁰ The Ministry of Labour reported that its “broad position” was therefore that women were as of now “eligible equally with men for all the training,” and “as regards the actual curriculum of training, the same course is available to men and women.” In the midst of crisis, the ministry conceded that “it has been found that women prove to be equally apt pupils,” even though in peacetime women would not have been trained for many of these jobs.²¹ Women were now allowed into all general engineering courses, including draftsmanship, fitting, inspection, instrument making, panel beating, sheet metal working, welding, electrical installation, and many more. Yet in the same memorandum that stated women were “equally apt” trainees, the Ministry of Labour also noted that only women who “show[ed] promise” would be “treated in the same way as men” and retained for longer training courses. The rest were given an abbreviated training period of a few weeks and sent out to work. Only trainees who were seen as having exceptional potential could rise above the social stigma that positioned women’s paid labor as simply a temporary stage in their lives.

As Ministry of Labour officials scrambled to keep up with war demands, they tried to make the most of available labor pool by increasingly adapting to the living patterns of married women and women with children. The ministry urged factories to set up training programs for women who could only work part time. Their labor was important enough to the war effort to warrant the waste and logistical inconvenience of rearranging training schedules and configuring shortened shifts. Creative solutions to the manpower shortage had to be sought for munitions production in particular.²² Much more so than it ever had in peacetime, the government recognized and tried to mitigate the “double shift” effect that accrued for married women who ran their households in addition to working outside the home, in order to keep wartime factories running efficiently and keep factory floor accidents down.²³

In the first months of 1942, over 150,000 women flowed into industrial placements each month. The Armed Forces survey for the year ending in June projected that over 700,000 men and 350,000 women were needed. In addition, Civil Defense and home front services such as nursing, the Women’s Land Army (farming), and the Navy, Army, and Air Force Institutes (NAAFI, which provided cafeterias and laundry services) required another one hundred thousand women. Munitions factories gobbled up over 350,000 men and 650,000 women. Altogether, this meant the armed forces and munitions industries utilized more than 1.1 million men and more than 1.1 million women during 1942.²⁴ Adjusting for the close to three-quarters of a million men in these statistics who were sent to fight in this time period, one can see that women workers held a sizable majority on the home front by the end of the year. The Ministry of Labour estimated that 80 percent of all single women between fourteen and fifty-nine years of age, 41 percent of wives and widows, and 13 percent of mothers with children under the age of fourteen were at work, or in uniform with the auxiliary forces, by 1942.²⁵

From there, their numbers only rose. By 1943, 90 percent of single women and 80 percent of married women between the ages of twenty and thirty-one worked for the war effort.²⁶ At the same time, any woman who already held a civilian job was legally bound to stay in it unless given explicit government consent to leave it. Two main categories of war work—an assigned job in a war-related industry or conscription into the Women’s Auxiliary Services—defined women’s experiences. The former could be any civilian job, many of which were in manufacturing. Jobs in the latter category often took the form of information work, from logistics, to data gathering, to military data-processing work.

From Listening to Morse Code to Breaking Enigma: Codebreaking’s Layers of Data Processing

In December 1943, a British team at the Post Office Research Station in Dollis Hill, London, finished assembling the world’s first digital, programmable, electronic computer under the direction of Tommy Flowers, a Post Office research engineer. After testing, they disassembled it to send it to Bletchley Park, where machine codebreaking operations using hundreds of much slower electromechanical machines—called Bombes—had been successfully underway since March of 1940.²⁷ In popular narratives of the war, men and machines formed an unstoppable cybernetic system on the battlefield, fed by information unlocked by top cryptanalysts. In reality, this relationship was far less neat and far more mediated by layers of communication and data-processing.

Over ten thousand women and men worked at Bletchley Park during the war and, like most home front industries, the workforce was overwhelmingly women.²⁸ For decades the substance of the key roles they played was overlooked or misunderstood.²⁹ Secrecy and national security meant Bletchley workers could not discuss their work either during or after the war: “We would wear no category badges, and if anyone asked us what we did, we were to say that we were secretaries,” recalled Eleanor Ireland, one of the hundreds of WRNS members who worked in codebreaking.³⁰ Today, many of these workers are unknown, having gone to their graves without ever talking about the content of their war work. Many of their identities—and even knowledge of entire categories of jobs they performed—have been lost.³¹

Reconstructing the chain of data-processing labor at Bletchley begins with the intercept stations that gathered encrypted data. The information provided to codebreakers like Turing and his colleagues was intercepted, recorded, read, and punched by hundreds of young women working as wireless operators and wireless transmission “slip readers” at the Knockholt intercept station in Kent before being transferred via teleprinter to Bletchley. The volume of traffic intercepted by Knockholt was so great that less than 10 percent of all intercepted messages were decrypted.³² This grueling and tedious work required extreme accuracy, and women would listen and transcribe encoded German wireless signals sent out in Morse code for hours on end in real time. “They would be transmitting all over the place and we would really have to cramp our fingers to write it down nonstop, because you only had one chance to get it,” recalled Joan Nicholls, who finagled her way into the women’s army auxiliary force—the Auxiliary Territorial Service (ATS)—at the age of 15 by claiming she was 17.³³

Wireless intercept operators had to record signals that seemed entirely random—the encrypted messages they listened to did not form words or sentences. Yet they listened so carefully that they got a feel for the grammar of these unintelligible signals and many intercept workers could recognize each individual German operator simply by the particular way he sent the Morse code—the particular staccato or lilt of the dots and dashes. It was essential to be “absolutely accurate” even to the point of knowing how many blocks of information one had missed if the signal faded. It “didn’t matter how tired you were or how sleepy or bored you felt, the minute that station came alive again you would be ... tearing pieces of paper off the pad and scribbling away like mad.”³⁴

The entire chain of information gathering and processing—from reading the intercepted codes and transferring them to punched tape with exacting accuracy to operating the electromechanical and later electronic computers that helped crack them—fully relied on women’s labor. Most of the women working at Bletchley came from the Women’s Royal Naval Service (WRNS) and were commonly referred to as wrens. The nickname was both a transliteration of the acronym for the Women’s Royal Naval Service and a pun on the British slang term bird—a slightly condescending term used to refer to women, especially young, single women.³⁵ Many of the WRNS members fit these categories: “96 percent of those who came were between the ages of seventeen and a half and twenty,” noted The General Report on Tunny, “though a few of the earlier wrens were rather older and more experienced.”³⁶ The WRNS members were given bunk beds in an old estate house, which Eleanor Ireland—an operator of the second, larger Colossus that broke codes for the D-Day landing—described as “very bleak, and very bare.”³⁷

Over the course of the war, these workers’ numbers rose dramatically. Max Newman’s machine codebreaking section, dubbed the Newmanry, was Bletchley’s mechanized heart. Here, the WRNS used the most advanced technology at the park. The Newmanry contained only two cryptanalysts and sixteen members of the WRNS in April 1943, but by April 1945, there were a total of 325 workers, including 273 WRNS workers (all women), twenty-eight maintenance and construction engineers (mostly or all men), and twenty-four cryptographers and administrators (mostly men).³⁸ At any given time, the number of women working far exceeded the number of men at Bletchley—on average, over 80 percent of the workforce was female. They fell into precisely the same demographic as prewar machine operators: young, single, white, and middle class.³⁹

The WRNS members used a number of different machines for codebreaking, and methods varied depending on the cypher. Early in the war, electromechanical computers formed the basis for machine codebreaking operations that attacked the Enigma encryption. These machines, known as Bombes, were designed by Alan Turing and Gordon Welchman who reverse engineered a captured Enigma machine. Manufactured for the government by the British Tabulating Machine company in Letchworth, Hertfordshire,⁴⁰ Bombes reproduced the action of eight Enigma encryption machines at once, with each rotor spun by an electric motor to quickly step through possible code settings.⁴¹ After setting up and running the machines, WRNS operators searched for combinations that were not logically inconsistent. They would then examine the promising output batches further to see if a longer string of code deciphered with the current setting produced German. By the war’s end, the WRNS members were operating two hundred of the huge, noisy Bombes, whose operation sounded like a mixture of loud clunking, thumping, and metallic scratching.⁴² Each one stood nearly six feet high and took up roughly thirty-two square feet of floor space.

Because the Bombes were electromechanical, however, their speeds could not match the increasing volume of critical encoded traffic. Encrypted information needed to be unlocked quickly enough for the Allied armed forces to be able to act on it. In addition, the relatively simpler Enigma code—which the Bombes helped crack—was being eclipsed by more complex encodings. Tunny (the British codename given to the German Lorenz code used between 1941 and the war’s end for top-priority messages from the German High Command) had far stronger encryption than Enigma code.⁴³ The machine used to encode Tunny messages used twelve wheels to set the code, instead of the Enigma machines’ three or four. As a result, breaking Tunny was far more difficult, time consuming, and labor intensive. A prevalence of certain letters in certain languages, along with attempts to guess “cribs” of likely text, had guided early breaks, but these methods were too slow and uneven to provide a long-term solution.

A depth, or a pair of messages sent using the same encryption settings, provided Bletchley with many early advantages, but as the war proceeded German forces enhanced security and the depths dried up.⁴⁴ As a result, a “statistical method” for brute-forcing analysis of encoded traffic, devised by codebreaker William Tutte, began to emerge as the only sure option. But Tutte’s method could take hundreds of years for a single message to be decoded by hand. It required more advanced machines, far faster than the Bombes, in order to work. Although Turing has gone down in history as the key hero of British codebreaking owing to his brilliance, his socially provocative and courageous political stances, and his tragic death, the machines and methods designed by Max Newman, Tommy Flowers, and William Tutte were more instrumental to the codebreaking that allowed the British to weather the later, more brutal parts of the war. These methods also ensured the success of the D-Day landings, which turned the war’s tide in the Allies’ favor.⁴⁵

The first machine effectively applied to Tunny was designed by Max Newman and deployed in June 1943. Partially electronic, but mainly relay based, this machine could compare two high-speed tapes of encoded information using a photoelectric reader. Unfortunately, it was neither fast enough nor reliable enough; the same data fed through twice might produce a different result each time. The WRNS operators who struggled to make it work named it the Heath Robinson after the rickety and comically pointless machinery of the cartoonist Heath Robinson—the British analogue of Rube Goldberg.⁴⁶ It was a nickname borne of profound frustration with substandard tools that did not inspire confidence.

Figure 1.2 A drawing of a fanciful machine (above) for “de-spiking” cacti by cartoonist William Heath Robinson (1872–1944) bears a striking resemblance to the codebreaking machine dubbed the Heath Robinson (facing page) by WRNS operators. Cartoon from Heath Robinson and K. R. G. Browne, *How to Make a Garden Grow* (London: Hutchison, 1938); photo of Super Robinson from I. J. Good, D. Michie, and G. Timms, *General Report on Tunny with Emphasis on Statistical Methods*, 1945, 382, HW 25/5, TNA. Courtesy of the Heath Robinson Museum, UK, and UK National Archives, respectively.

If the Heath Robinson could perform its analysis successfully, the resulting data would be run through the Tunny machine, an electromechanical machine that had been created by reverse-engineering the German Lorenz enciphering machine. The WRNS recruits performed this analysis on each encoded message in order to find encryption settings for that day’s messages. Working with the machines designed to break Tunny code consisted of trying different wheel settings and repeatedly resetting the plug boards with settings taken from the runs on the Heath Robinson machine.⁴⁷ “Tunny was plugged up just like an old fashioned telephone exchange,” explained Dorothy Du Boisson, an early Colossus operator. She added, “It used to give me electric shocks as I put in the plugs.”⁴⁸ Tunny machines also had a paper tape reader to take input from the Robinson machine, but the two Heath Robinson machines perpetually broke down, mangling the tapes that were to be fed into the Tunny in the process.

As a result, operators had to perfect the art of how to keep the tapes intact through the Heath Robinson’s high-speed runs, which often stressed the paper media to the breaking point. As a last resort, glue was used to repair tapes when they broke, even though pieces of tape sometimes flew all over the room if they had been spinning at a high speed. The WRNS workers would collect the pieces and meticulously put them back together. “We went into a very huge room where they produced the tapes and repaired them. We had Bostick, and we used to Bostick them together,” recalled Eleanor Ireland.⁴⁹ Even when they did not break, stretching of the tapes could produce erroneous results. Handling the input and output of the machine, in addition to designing their runs and programming them, was a delicate and complicated process requiring both skill and hard-won tacit knowledge. WRNS members were all too aware that the outcome of their work could mean the difference between life and death for British soldiers serving at the front, as well as for the civilians being bombed on the home front. Given that these young women had family and friends in the line of fire, the idiosyncracies of their machines were as frightening as they were frustrating.

Max Newman, who was insulated from the problems of running the machinery that made up his unit, requested more Heath Robinson machines be built in order to speed codebreaking. However, the solution to the problems of these unwieldy and inaccurate machines was not simply more of the same. In the opinion of leading Post Office engineer Tommy Flowers, the only way to speed the pace of codebreaking further—and enhance accuracy—was to use an all-electronic computer. This presented major problems, however, because it had never been done before and there was no time for a proof of concept. In the United States, the ENIAC project was barely off the ground. Indeed, ENIAC would not be functional, nor be public knowledge, until after the war, and its uptime for computing would not be reliable and productive until well after that.⁵⁰ In addition, ENIAC was focused on completely different computing objectives: It was designed for mathematical computation for ballistics trajectories, not logical programming required to break codes.

Flowers faced an uphill battle to get approval for his project. Aside from the Heath Robinsons, only electromechanical machines like the Bombes had been used consistently for codebreaking success on both sides of the Atlantic. Newman was sure a machine with thousands of vacuum tubes would not work reliably. Statistically, he reasoned, some of the tubes would be failing at nearly every given moment, leading to perpetual downtime and a machine that was effectively unusable. Plus, owing to wartime shortages, the vacuum tubes and other supplies needed to construct such a machine were scarce—as was money. Although Flowers was eventually given enormous leeway and resources, he later revealed that he also spent significant amounts of his own money in order to complete the project.⁵¹

Wrens Shoulder a Colossus: Britain’s Secret Technological Revolution

In late 1943, Flowers and his team completed a behemoth of his own design that contained well over one thousand vacuum tubes.⁵² Dubbed the Colossus, it took Newman and his machine codebreaking section by surprise.⁵³ It manipulated data electronically, obviating the need to compare two high-speed punched tapes.⁵⁴ The data was still fed into the machine using high-speed punched tape, however, contributing to an extremely noisy working environment that was a far cry from later electronic computing installations. By running a lower current than normal through the machine’s many vacuum tubes, and never turning the machine off (since most tubes failed during the stresses of warming up or cooling down), Flowers was able to ensure a level of consistent, reliable uptime that would be unheard of in other computer installations until well into the 1950s. With this machine, Tutte’s statistical method, which was essentially a way of brute-forcing codebreaking by trying a huge number of possible solutions at extremely high speed, suddenly became a reality. Colossus allowed a method that had been theoretically sound but, up until this point, impossible in practice to become the best tactic available for breaking Lorenz code. Immediately after seeing the first Colossus’s capabilities, Newman asked for more. By the war’s end, there were ten of the massive, room-sized Colossus computers at Bletchley, operated around the clock by close to three hundred WRNS recruits.

The WRNS programmed Colossus using switches and plug boards in order to electronically emulate the different wheel and pin settings of the encryption machines. When Colossus arrived in January 1944, it took over the role of finding the patterns and wheel settings for the Tunny machine from the Heath Robinson.⁵⁵ Able to process twenty-five thousand characters per second, Colossus I and II provided critical, nearly real-time intelligence for the D-Day landings. Colossus II in particular was instrumental in allowing the Allies to get the information they needed to launch a successful offensive: The information it decoded allowed Allied forces to see that the German Army was moving away from the Allies’ planned landing site, having been fooled by intentionally leaked information that said Allied armies would be elsewhere. Yet Colossus II was not a smoothly working upgrade that effortlessly expanded the Allies’ information arsenal. It was installed less than a week before D-Day, under extreme pressure and suboptimal circumstances. Flowers’s team worked nonstop to get it operational. Once it was, the operators had to wear thick rubber boots so as not to be electrocuted by the machine, because the machine room floor had flooded.⁵⁶

Because of the complexity of the encrypted code, the WRNS operators used each Colossus to break a different level of encryption. Operators had to repeatedly set and reset the machines and manage paper tape input and output for each run. The harried pace and makeshift machine rooms contributed to a frenetic environment in which workers had to contend with high levels of noise and improper ventilation even before the installation of the Colossus computers.⁵⁷ When Colossus came, these problems were magnified: The computer made a significant amount of noise owing to the high-speed paper tape input. Performing precise work under extreme time pressure was particularly arduous in such surroundings. Since each Colossus had well more than a thousand vacuum tubes, which were never switched off in order to preserve their longevity, they significantly raised temperatures in the non-air-conditioned machine rooms. While cooling off with coworkers outside the boiling hot Colossus room one night, Dorothy Du Boisson (pictured in figure 1.3, on the left) recalled a male coworker suggested she and her fellow operators should go topless. But, she added dryly, “we did not take up the offer.”⁵⁸ In their off hours, contaminated water, poor heating, and noisy living quarters heightened the stress for WRNS workers. In addition, because WRNS supervisors outside the park were unaware of the skilled, nonphysical nature of the work being done, they demanded mandatory morning drilling, which resulted in exhaustion and illness for many workers until their supervising officers were eventually told to stop the drills.⁵⁹

Figure 1.3 WRNS members working on a Colossus. The workers have been identified as Dorothy Du Boisson (left) and Elsie Booker by historian B. Jack Copeland. I. J. Good, D. Michie, and G. Timms, *General Report on Tunny with Emphasis on Statistical Methods*, 1945. Courtesy of the UK National Archives.

Like many other women in Bletchley’s data-processing edifice, Colossus operators worked around the clock, in three eight-hour shifts, alternating shifts on a weekly basis. Each woman was allowed to leave only one weekend a month. Many worked extra hours, to the point of exhaustion, especially during times of exceptional need. These difficulties were shared by all WRNS members and indeed all Bletchley staff, not just those working in the Newmanry. At one point, the entire teleprinter transmission staff worked twenty-four hours straight without any break to keep up with the volume of intercepted messages.⁶⁰

A typical work shift in the codebreaking section utilized seven cryptographers and sixty-seven WRNS workers, with one or two of the men cryptographers supervising. (There was only one cryptographer at Bletchley who was a woman: Joan Clarke. She is perhaps best known, despite her accomplishments, for being Alan Turing’s “beard” for a short time.) Most of the WRNS members on a shift would work as Tunny or Colossus operators, about twenty for each type of machine, while two would operate Robinson machines and seven functioned as registrars. Because of the secrecy surrounding British wartime intelligence, the exact number of people involved with the Colossus computers and their precise roles may never be known. In 2014, Joanna Chorley, a former Colossus operator, found perhaps the only remaining photograph of thirty-nine of the operators who were present at Bletchley near the end of the war.⁶¹ Another woman, Margaret Bullen, came forward the same year to talk about how she and other women wired up one of the Colossus machines and soldered together its connections in order to prepare it for the WRNS to use it, showing that despite previous assumptions women did perform computer maintenance and engineering work at the park.⁶²

In the machine rooms, the women who worked as registrars were critical to successful, efficient codebreaking. Their job was to record and schedule the computers’ work assignments so that the problems to be solved were arranged in order of priority and also would flow onto the machines at a constant rate to maximize the available computer power. The registrars had some of the most important work at the park—they needed to quickly solve ever-changing logistical and operational problems under extreme time pressure. While the “details of Tunny Room and Colossus jobs were left to the operators concerned,” the work was orchestrated by the registrars and “returned to them on completion,” noted the definitive report on operating procedures.⁶³ The registrars called the shots, even though the WRNS members who operated the machines had a significant amount of responsibility and autonomy. Everyone at Bletchley felt their responsibility acutely, knowing that work slowdowns translated into more civilian and soldier deaths. Hard-won breakthroughs made a huge difference to morale: “There was a great deal of jubilation when we broke one particularly important link,” recalled Du Boisson, adding that “to celebrate, someone decorated the room with daffodils.”⁶⁴

Killing the Golden Geese

Churchill famously described Bletchley’s workers as the “geese who laid the golden eggs but never cackled.”⁶⁵ His metaphor contained an implicit reference to the WRNS—the dutiful “birds” in the nest of the nation’s codebreaking operations. But narratives of women’s heroism nonetheless downplayed or obscured the levels of skill and training required for their work. Far from being deskilled functionaries, the WRNS were essential to the nation’s cryptanalysis and codebreaking efforts.

The official report of the work at Bletchley Park, kept secret until 2000, stated that several members of the WRNS “showed ability in cryptographic work.”⁶⁶ Despite this, none were promoted to cryptanalyst positions.⁶⁷ Most possessed what was somewhat condescendingly described as “cheerful common sense,” but some were trained to test the machines, inaugurating them into a different level of work responsibility from the role of operator.⁶⁸ Prior to testing the Colossus computers, WRNS workers were also responsible for testing the majority of the Bombes.⁶⁹

Indeed, all operators needed to understand the Tunny machine and Colossi well enough to be able to fix minor breakdowns, both for reasons of convenience and because of the shortage of official maintenance engineers, all of whom were men. The male-oriented nature of apprenticeship to engineering maintenance work created a dire shortage of maintenance engineers for Bletchley’s machines. As a result, a special nationwide program of recruitment for engineering trainees was established for sixteen- to nineteen-year-old boys. These very young men with no experience were seen as a better investment than women who already worked with the machines because they were expected to have much longer careers, justifying the training involved.⁷⁰ Innate ability was not a factor. Throughout the war, supervisors struggled to decide on how much to gender jobs in the midst of crisis; an April 1942 report from Bletchley Park investigated the “number of new posts that can be filled by women” and stated the percentages women should take up for each job class.⁷¹ Hand and machine computing work formed the majority of the posts deemed suitable.

The official report of Bletchley’s codebreaking operations stated that “wrens (unlike men) were organized in fixed watches and given fixed jobs in which they could become technically proficient,” but the accounts of the WRNS members themselves differed from this characterization.⁷² “We had to be versatile,” said Du Boisson. Initially operating Colossus under the direction of a cryptographer who told them the settings required for the run, the WRNS members soon took responsibility for finding the settings themselves, “freeing the cryptographer for more important work.” Another Colossus operator recalled that she was taught to operate the machine by a fellow WRNS member rather than a supervisor.⁷³

Training agendas at Bletchley reflected the fact that the machine work conducted by women was central to codebreaking. Arriving members of the WRNS were given two weeks training in binary math, the teleprinter alphabet, sight-reading punched paper tapes, and the structure and workings of the Tunny and Colossus machines.⁷⁴ One Bletchley codebreaker recalled Bletchley as a sort of “high-pressure academy.”⁷⁵ Chosen for their posts by interview, all WRNS workers had passed secondary school mathematics exams, and 9 percent of them had been to university. Over one-fifth had a Higher Schools Certificate, 22 percent had had some previous training, and 28 percent had previous work experience.⁷⁶

WRNS members were also included in the park-wide program of further education for all workers. Printed and oral lectures were organized by a team of one or two cryptanalysts and several members of the WRNS for both men and women working at Bletchley. “It was the policy of the [machine codebreaking] section that all its members should be encouraged to interest themselves in all its activities and to improve their theoretical knowledge,” the official report on Colossus noted, because “in practice it became increasingly hard for wrens to get a complete picture of an organization in which they might only have done one job.”⁷⁷ By 1945, regular lectures aided WRNS workers in becoming better codebreakers as the war drew into its final, grueling stage. These theoretical, mathematical lectures and seminars were “a complete success” for the Colossus operators. The same training was given to new young men recruited for the ostensibly higher-level codebreaking work that did not involve machines.⁷⁸ The mounting exigencies of war increasingly destroyed the logic behind gendered training, especially as women outnumbered men by an ever wider margin.

Interestingly, men who supervised machine work were there less to facilitate the work or give guidance to the operators, and more for their own benefit. Park leaders felt male cryptographers came up with better theoretical insights when they alternated their cryptographic research with practical, applied work.⁷⁹ Machine work—and the theory and skills it required—was an integral component both intellectually and functionally of the codebreaking process. It was not, as many assumed due to its feminized nature, deskilled drudge work. Supervisory cryptanalysts worked hard to maintain an atmosphere of egalitarian teamwork. One WRNS member who worked under Max Newman recalled that on arrival he decreed everyone working under him should call each other by their first names.⁸⁰

Intelligence jobs dominated by women that were positioned as peripheral to the main activities of Bletchley—like the high-speed slip readers at other installations who copied enciphered traffic from radio transmissions in real time—were also skilled and, during wartime at least, recognized as “highly trained.”⁸¹ In a particularly striking example, the “speed of decoding” in the section headed by Major Twin was more than three times slower simply due to a lack of skilled typists: “A Hut 8 girl can do in 2 hours what Twin can decode in 6–8 hours—on one day traffic took 18 hours.”⁸² Although women workers like these helped prove the value and ensure the success of the earliest electronic computing systems, secrecy surrounding wartime codebreaking operations often submerged these narratives and hid their importance.⁸³

Despite the importance of the work they did, machine codebreakers often did not even know which arm of the German military had produced the codes they broke.⁸⁴ Workers at Bletchley could only guess what was going on in other sections of the park since projects were atomized for security. They pieced together a picture of the larger military strategy from odds and ends of gleaned information. Secrecy was taken so seriously that many Bletchley machine operators remained silent for decades. When issued her security pass on arrival, Eleanor Ireland recalled that she was instructed to protect it with her life.⁸⁵ “We came out of there fairly traumatized by the whole thing,” she recalled.⁸⁶ Forbidden to speak about their work or location under penalty of imprisonment, no one at the park was allowed to ask anyone in another Bletchley unit what he or she did.⁸⁷ Colossus operator Catherine Caughey lamented, “My great sadness is that my beloved husband died in 1975 without knowing what I did in the war.”⁸⁸ Even Tommy Flowers could make little use of his wartime electronics experience because of the secrecy surrounding his successes.

These extreme security measures continued to influence the historiography of labor and technology for decades after the war, paradoxically ensuring that British accomplishments went down in history as also-rans in a US-centric story of early electronic computing. One high-speed wireless operator and Morse slip reader in the telecommunications wing of Bletchley worked only “steps” away from the Bombes and Colossi, yet like most of the world she did not know of the machines’ existence until bits of information began to trickle into the British press in the late seventies.⁸⁹ Conversely, the codebreakers knew nothing about the vital information transmission that allowed them to break codes. Even well into the nineties, the curators of the Bletchley Park museum were unaware of the existence of this wireless operator’s telecommunications unit or the critical nature of this data transmission work in the codebreaking information chain.⁹⁰

Notwithstanding their abilities and their experiences as the first electronic computing workforce in the world, most of these workers were made redundant at the war’s end. One woman from the wireless transmission section at Bletchley tried to make a career in industry after VE day. Having been given over a year’s training as a high-speed wireless operator while in the Women’s Auxiliary Air Force, she was well qualified. Yet her plans evaporated when she found, to her surprise, that the main cable company was now only accepting men.⁹¹

The operators of the Colossus computers likewise dispersed to other jobs for the most part, according to available records. Most operators went on to work in fields that were unrelated or only tangentially related to their pioneering computing work, though not for lack of interest. After being ordered to break up their machines into pieces “no bigger than a man’s fist” by Churchill’s edict at war’s end, the WRNS workers were made to again sign the Official Secrets Act and sternly ordered to remain silent. Refusing to accept the total erasure of their work, some secretly snatched mementos from the wreckage. Eleanor Ireland remembered grabbing and concealing a “little, blue” vacuum tube that she treasured for years afterward.⁹² At the same time, two Colossus computers survived and were transferred to GCHQ for top secret use throughout the 1950s and into the 1960s. Although records indicating what personnel may have transferred with them are scarce or in some cases still closed to researchers, it is reasonable to infer that at least some WRNS members helped set up the machines at GCHQ. The tacit knowledge built up by the WRNS during the war had become a necessary component in this computing system—no one else knew the practical ins and outs of programming and operation for the machines as well as they did. Even Tommy Flowers had destroyed his schematics of the machines as ordered, meaning that the secret knowledge contained in workers’ heads took on an even greater value.

Questions about Colossus operators’ postwar roles aside, however, not all of Bletchley’s women information workers were discarded or submerged by government secrecy at the end of the war. The fact that the experiences of women shunted out of the field at war’s end have expanded to define the texture of women war workers’ treatment obscures the important continuities and connections between wartime and peacetime information technology systems and the workers who ran them. In fact, the focus on the Colossus computers runs the risk of rendering invisible many of the other continuities in computing after the war.

The Girls in the Machine

For the most part, the end of the conflict meant a loss of status and opportunity for the many women who had become a critical component of the nation’s information infrastructure. Postwar unemployment for women was worsened by “a recurring demand that such jobs as were available should be reserved for men, and women should go back to the home, whatever that might mean,” observed the head of the TUC Women’s Conference after the war.⁹³ The temporary reprieve from many gendered job classifications given to women by the government’s conscription program was now over, but this was not a key factor in computing work, which had been gendered feminine even before the war.

As a state-building tool in times of stability and peace, as well as in times of war, computerization’s history is largely a narrative of how the data-intensive state expands its reach and power by taking for granted, and rendering invisible, ever-greater numbers of information workers.⁹⁴ The work itself was meanwhile constructed as feminized and deskilled despite its apparent complexity. When information about the Colossus computers began to become public in the 1970s after being kept secret for decades, women’s work with these computers was assumed to be low level and lacking in any significant skill and responsibility. Conceived of as a passive support role to the “real” work of codebreaking, rather than an integral part of the codebreaking itself, women operators paradoxically found themselves and their work undervalued in histories that were being rewritten to privilege the role of the computers they ran. The lower esteem in which women’s labor had long been held resulted not only in unequal opportunities and pay, but also in the perception that the work being done by women was somehow implicitly lower in skill and importance. As Annie Burman has noted in her study of gender segregation in codebreaking work at Bletchley, “the discursive conviction of women’s inferiority” and the ways in which women’s work was discussed using terms that removed their agency and expertise, even down to the practice of calling them girls instead of women, has made it easy to “fall into the trap of assuming that systems built on degrading jobs given to women” actually made these jobs less important.⁹⁵ This remained the case even in situations where the work was materially similar or completely identical to work done by men. The double-edged sword of feminization both helped and hurt the progress of computing early on, as it defined understandings of how to structure and deploy large-scale computational projects.

Before the war, the most complex and important administrative and accounting work of the government was handled by room-sized punched card installations. From payroll and pensions to the compilation of census statistics, punched card systems defined how much data the state could manipulate and how quickly and well it could produce and utilize this new information. These systems—made up not only of machines, but also of workers, expertise, and finely tuned processes—redefined the government’s scope, creating the capabilities of the modern state. The expansive, data-reliant economic and social programs that characterized modern British and Western European states grew in tandem with computing capabilities. Nationalized healthcare systems, social welfare systems, and the complicated taxation regimes that supported them—not to mention the collection, storage, and manipulation of census data required to ascertain how to provision these resources—all required steadily increasing data-processing power. During wartime, increasingly automatic, systematized information processing saved the state; during peacetime, it expanded it.

Yet after the war, women’s war work was firmly constructed as Taylorized and skill-less—a discourse that began even during the conflict. One letter writer to the Times said what most felt when he called women “runs of the industrial commonplace” who were only able to work in war manufacturing because the processes had been broken down into smaller, simpler pieces.⁹⁶ For evidence regarding the recent public perception of women’s wartime computing work, one need look no further than the London Science Museum’s 2012 exhibit on wartime codebreaking. The museum descriptions state that Bletchley’s “machines operated around the clock.” Rather than saying who operated the machines around the clock, the exhibit removed the presence and agency of the operators and made the machines seem far more automatic than they actually were. Another description states that women “tended” the machines, rather than more accurately stating that they operated, troubleshot, tested, programmed, repaired, and even helped build them.⁹⁷ In a similar example of erasure, the two WRNS members in the only surviving picture of a Colossus being operated are not named in any of the exhibits at the UK’s Bletchley Park Historical Site and National Computing Museum, although their identities are known.⁹⁸

Press accounts of the time made sense of women’s wartime accomplishments by using the conceptual umbrella of the “home front,” which downplayed the major dangers and privations involved. Over fifty thousand Britons were killed on the home front during aerial bombing raids. For 1940 and 1941, the years of heaviest civilian casualties, civilian deaths accounted for nearly one-third of all wartime deaths, and during the height of the Blitz more casualties were recorded in Britain than for British forces deployed elsewhere.⁹⁹ The many women who sustained wartime injuries and disabilities received lower payments from the government than men.¹⁰⁰ Their treatment reflected a long-standing, ostensibly commonsense notion that women needed less and contributed less in work and public life.

At the same time, automating technologies increasingly deployed in offices during the twentieth century were creating a postindustrial labor landscape firmly divided upon gendered lines. As British feminists used the war to advocate for equal pay and equal treatment, the developing technological infrastructure organizing peacetime data processing worked to subtly undercut their efforts. The process of further computerization reflected and helped strengthen patterns and expectations that demanded women’s labor be seen as secondary or tertiary within the very technological hierarchies where it was a foundational element.

In 1944, the wartime Womanpower Committee and affiliated groups of women workers mounted a major push for equal pay in the Civil Service as a result of the war.¹⁰¹ The thin end of the wedge was asking for payment of equal injury claims to women who had sustained war injuries. Because women were paid only a percentage of men’s wages, the government likewise only gave them a percentage of the men’s payment for wartime injuries. The committee sought to change this, but gendered economic logic of this kind was widespread and ingrained. Indeed, it structured the economy, because government and industry took unequal pay scales as a precedent and guide for other payments—from injury claims to pensions.

Key to attacking unequal pay was changing the terms of women’s labor from being temporary and supplementary. The repeal of the marriage bar—the rule banning married women from keeping their jobs in the Civil Service (the nation’s largest employer), nationalized industries, and as teachers—was a critical first step. Separate women’s pay scales throughout industry and government continued to economically reinforce assumptions about women’s dependence. Equal pay campaigners, working since before World War I, were unable to give the government enough compelling reasons to enact a moral abstraction of equality, which many assumed would be economically and culturally disadvantageous in practice. As a result, the war left the economic and legal structures that institutionalized women’s dependence and their limited opportunities in the labor market largely intact.

Even in clerical and administrative work, a steadily feminizing category, women workers lost out. A clear example was the fate of women civil servants after the war, because “perhaps as much as 50 percent to 100 percent of the established [permanent] complements” had been replaced by temporary women workers due to emergency hiring during the war and postwar reorganization.¹⁰² Many wartime women temps lost their jobs after the war, because the government would only agree to retain temporary women in 15 percent of all vacancies in the postwar reconstruction of the service. Although the staff negotiated for a general examination to allot the permanent posts—after all, the Civil Service was an exam-based meritocracy—the government refused. Instead, the responsibility of choosing candidates was given to the department heads, a system that powerfully reinstilled cultural biases into the structure of the state’s institutions.¹⁰³

As a head of the TUC Women’s Conference observed after the war, unemployment for women was worsened by the end of hostilities because “women were regarded as a sort of auxiliary force which, to use a military term, was expendable in time of crisis.” This created a topsy-turvy world for women workers.¹⁰⁴ For them, the crisis period in which they became expendable and had to struggle to find work was peacetime. Women trade unionists bristled at the prospect of losing opportunities afforded by World War II, and many other women longed for the opportunities that war brought. Mary Coombs, perhaps the first woman to work in commercial, rather than government, computer programming, recalled that she “half hoped the war would last long enough to join up” when she was in school.¹⁰⁵ Much of women’s wartime training became wasted—not because they chose to drop out of the workforce, but because they were no longer considered good candidates to hire when men returned to the labor pool.

Only in certain jobs—ones that both retained a feminized image and suffered labor shortages—did women have a fighting chance. The most important exception to this lack of labor continuity for women as the war ended was within administrative and scientific computing, fields where women still retained their status as the favored labor pool. This exception seems on the surface to be a simple continuity from before the war, but in fact it was an important outlier in a context where most women—even those in feminized jobs like clerical work—found themselves fired or struggling to stay employed at the war’s end. The gender continuity in the realm of nonelectronic computing was emblematic of a larger changeover to a postindustrial information economy in the second half of the twentieth century, a time when women’s labor would play an ever greater but paradoxically more hidden and devalued role.

Although the Colossus operators dispersed after destroying their machines, the many “local girls” hired to run the Hollerith equipment at Bletchley alongside the WRNS members—the “girls” who not only operated but also programmed the machines using plugboards after writing up the programs by hand—did not disperse and disappear. Instead, they went on to work in jobs of similar complexity and content in the same field: “At the end of the war with the setting up of GCHQ it was proposed that a Hollerith installation should be included in the organization, together with the personnel who had worked under Freeborn at Bletchley, and who might wish to continue in this type of work,” wrote Ronald Whelan, the head of the Hollerith installation.¹⁰⁶

The Government Communications Headquarters (GCHQ), known as the Government Code and Cipher School until the end of the war, had played a critical role in intercepting and collecting enemy military information to create actionable wartime intelligence. Therefore, the invitation extended to the Bletchley Hollerith operators to continue their work after the war, in the context of GCHQ, represented an incredible opportunity. Here, women could do relatively important work even after the cessation of hostilities. However, for this the staff needed to become civil servants, taking the exams for entry into the service’s complex, nominally meritocratic hierarchy. The women who had proven their worth doing this work during the war under extreme pressure and hardship still needed to prove their abilities according to the government’s peacetime standards. The women could compete for places in the Civil Service’s prestigious exam-based meritocracy, with no guarantee they would get the jobs. The dynamics of Civil Service hiring were complex and often used quotas to hold down the number of entrants in a given job class. “Providing those wishing to take this step were judged to be suitable” according to government standards, they would transition into secure, life-long jobs in the government’s service. It was with pleasure and relief that Whelan was able to report “that those who wished to do so” had been judged to be up to snuff, and “duly became civil servants.”¹⁰⁷

With that, an influx of several hundred workers transferred from wartime operations to peacetime government administration in one of the largest single blocks of wartime workers—and certainly the largest block of wartime intelligence workers—transitioning en masse into the peacetime economy. Although other women lost their jobs, these women were important enough to keep theirs, because they were harbingers of the shape of things to come. GCHQ’s “Hollerith girls” became part of a high-stakes, highly gendered technocracy, forged in the crucible of war, that would continue to define British government and society until the century’s end. They became some of the first and highest-level computing workers in a technological hierarchy that would use women exclusively to form the broad base of the information technology pyramid. These war workers, though not privy to the top-secret electronic computing work that had gone on during the war, represented the lion’s share of workers in computing at the time: women who worked with electromechanical calculating and data-handling systems.

Electromechanical systems would continue to define the shape of industry and government—and the possibilities of administrative computing for many years to come—even as electronic systems became commercially available. The content of this work was very similar, and the skills required for it virtually identical to what was required for early electronic computer installations in the 1950s. As a result, it was a training ground for early computer workers, and many women in jobs like these went on to operate and program electronic computers. That these computing workers were repositioned in the peacetime context of the state’s meritocratic bureaucracy as the war ended, specifically as a result of their war work, makes them a key example of how swords were turned to plowshares in computing. Although the technological disjuncture caused by the destruction of the Colossus computers is well known, the process by which wartime computing labor transitioned into the peacetime context is less well known.

The experiences of these women not only powerfully contradict the general assumption that all women left or were forced to leave their war-related jobs at the end of the war, but they also show how women were able to continue in computing specifically as a result of their war work. Their experiences and their machines—though less glamorous and groundbreaking than the Colossus operators and their faster, state-of-the-art machines—represent a clear continuity between wartime and peacetime computing labor and correct the misperception of electronic computing as somehow a separate class of work from electromechanical computing. In fact, there was significant continuity and convergence when it came to electromechanical and electronic computing labor.

Although women faced a postwar labor market in which employment opportunities were not of a similar level of complexity or responsibility, they did not disappear from computing with the return of men from the front, as is sometimes assumed, or with the destruction of the earliest productive electronic computing system, whose practical success they had helped ensure. The Hollerith operators remained as key participants in the emerging data-driven state. After the war, they became integrated into a system that would be no less complex, and far larger, than all wartime intelligence gathering.

As high-ranking civil servants retooled the technological infrastructure of the state for peacetime, an emerging technological regime based on ever-greater computing power would have a critical impact on women’s labor equality. Because peacetime computing in the service of the state would rely on strictly gendered labor categories even more than wartime computing had, women’s labor would become configured as supplemental rather than foundational, even as this labor continued to form the backbone of the country’s data-processing systems. So taken for granted were these technical women workers that they became seen as part of the computing system itself. As more massive, more complex, and faster data-processing systems took hold in government and industry, these workers were consolidated into an essential underclass of non-working-class, sub-white-collar office labor. This classification would have an impact on the long-term productive capacity of the nation as a whole in significant and unexpected ways.

Nowhere were the fraught gendered dimensions of the transition from a war to a peacetime economy—and the role these dimensions played in women’s lives—more apparent than in the Civil Service. The Civil Service would become ground zero for the changes in computing labor that would define women’s position in the new technology landscape as not only subordinate but as fundamentally separate from their male peers. Starting in the immediate postwar years, gendered labor reorganization in conjunction with technology defined the government’s vast workforce and the fiber of its institutions for decades to come. Yet because of the Civil Service’s unique position as a nominally meritocratic sphere under the government’s direct control, it was also the most logical place for labor reformers to try to address the pressing gendered labor inequalities that defined these jobs.

Equal Pay, Marriage Bars, and Dead-End Machine Work

Women’s growing numbers and their potential power in the white-collar labor market were fast becoming seen as one of the biggest threats to the national order since industrial labor organization. The government particularly feared that long-established labor hierarchies would be upended and that huge costs would ensue if the persistent campaign for equal pay in the public sector could not be squelched. Before equal pay, however, another issue stood in the way of postwar women’s worth: the marriage bar. In government as well as industry, women were required to leave their jobs upon marriage. Legally, the practice should have been abolished by the Sex Disqualification Removal Act of 1919, which aimed to eliminate exclusion from jobs on the basis of sex or a woman’s marital status.¹⁰⁸ The 1919 act was broad and ambitious, but it was unevenly applied and it contained no enforcement machinery. One of the earliest legal claims brought under the act, against the practice of firing married teachers, was denied outright.¹⁰⁹ Even aristocratic women were unable to use the law to their advantage: In 1922, the daughter of Lord Rhondda tried and failed to get permission take the place in the House of Lords bequeathed to her by her father.¹¹⁰

In white-collar professions, women made uneven gains; the legal profession and accountancy opened their doors to women at this point, but the Civil Service, the nation’s largest employer, continued to forbid married women from continuing their careers. Even though it had long been described as a “fair field with no favor,” the Civil Service required women to resign on marriage, thereby mandating inequality of opportunity.¹¹¹ Relaxed during World War II, the rule was swiftly reinstituted at war’s end. Baroness Beatrice Nancy Seear, a labor reformer who worked in government and industry, recalled that in the Civil Service at war’s end, “those women had to be out of the men’s departments at the speed of light,” even though there was no danger “that men wouldn’t get their jobs back.”

The 1919 act, along with the 1918 Representation of the People Act that gave the vote to a limited number of British women for the first time, seemed to come as a reward for women’s dedicated war work on the home front during World War I and seemed to be a move to preempt the revival of the violent and disruptive tactics of the prewar suffrage movement. Women labor reformers hoped that the end of hostilities in the most recent war and women’s wartime service could similarly work to their political advantage. Reformers like Seear saw “absolutely no excuse” for how women were made to leave their jobs after the war, noting: “It was just sheer status quo. And back these women had to go.”¹¹² Women’s paid work remained perceived as largely “incidental” rather than as economically or socially necessary.

This image clashed with the actual shape of the government’s work force: Women accounted for over 66 percent of the main Civil Service clerical union after World War II.¹¹³ Many married women did not receive an adequate wage through their husbands, and others were raised in a tradition of working, politically independent women. As the nation returned to prewar norms, the interests of women workers once again became submerged in trade unions and professional organizations dominated by men. Women thus had little power to present pay claims or other complaints to employers through their unions. Rather than continue to fight, observed Seear, many women “accepted a great deal of this. The national agreements had two rates, one for women and one for men ... They took it for granted.”¹¹⁴ In the aftermath of World War II, the government’s attitude had changed little from just after World War I, when a War Cabinet report stated: “Girls and women have regarded their work as incidental rather than as a main purpose of their lives.”¹¹⁵

This idea echoed in the government-commissioned Equal Pay Report of 1946, which stated that women were ancillary to the workings of the labor market: “It seems sensible to assume that virtually all men are and always have been and always will be in the market for employment. But experience shows that the proportion of women seeking employment is variable within limits in response to the influence of social forces” (italics mine).¹¹⁶

The attitudes of the Equal Pay Report and other similar government reports disciplined women into certain roles as much or more than they described women’s actual roles. As a young woman, feminist reformer Enid Hutchison was mentored by women who had been allowed to achieve relatively high positions in the Civil Service and in industry owing to the demographic shifts created by World War I. Seeing these possibilities, Hutchison determined not to go into nursing or teaching—the jobs usually “meted out” to educated women in the early twentieth century—after finishing her university degree.¹¹⁷ Raised in the weaving district of Lancashire, she was accustomed to a labor landscape in which working women only took time off temporarily to have children and did not end their careers upon marriage.

Although many women complied, willingly or not, with the regulation to resign their Civil Service posts upon marriage, a difficult-to-measure minority revolted. Some of these women got around the restriction by not obtaining a formal marriage license. Others used different kinds of subterfuge to try to travel a middle route. In bustling London, whose population peaked at nine million people after World War II, the anonymity of the crowd sometimes meant that women could hide their marriages.¹¹⁸ Hutchison recalled that “there was always this thing, that if you married, then you had to have babies surreptitiously, as people did, be married surreptitiously as people could do in the London area ... You had to live a life where you couldn’t really be known.”¹¹⁹ Women unable or unwilling to live this kind of secretive double life had few options. At age twenty-seven and “getting on in years,” Hutchison herself gave in and married her longtime companion, also a government employee, who was unable to hide the engagement due to his relatively high position. Hutchison’s supervisor was reluctant to lose a capable, trained employee, but she nonetheless fired Hutchinson in accordance with regulations when Hutchison refused to resign.¹²⁰

Like many other women, however, Hutchison did not stop working after losing her job. She had worked her way up from “the lowest possible job that it was possible to have in the Ministry of Labour, a temporary woman clerk III” (emphasis in the original) and now restarted the process in industry. When she was able to return to government service due to the labor shortages of World War II, she recalled with exasperation:

It was back again to the Temporary Woman III grade at the bottom of the heap. Once more you go to the bottom of the heap. ... I went to see the man in charge of the Assistance Board Offices, because I knew I could get a job straightaway there, really. And he said “The trouble is, you’re married” ... And I said: “does that still operate? Isn’t there a war on?” So he rang up his headquarters and he was told he could employ a married woman. Within a few months, all the staff were married women, and he was lucky to get them. But that was [still] the regulation. The regulation had not been changed!¹²¹

War’s end meant the return of the marriage bar. Given the large and growing proportion of civil servants who were women, this variation in treatment according to gender undercut the government service’s meritocracy in the broadest sense. Some women were allowed to come back to work in the Civil Service after being forced to resign, but they were only allowed to return as temporary workers, at the lowest pay and skill grade. It was usual for women who had worked their entire adult lives to find themselves downgraded to the lowest rung on the career ladder after marriage, with no pension rights or other benefits, in both the public and private sector. The marriage bar did not so much prevent women from working as it prevented career progression.

Machines as well as people played into the decision to return to the marriage bar after the war. The government feared that removing it would negatively impact large swaths of lower-level workers doing machine work, with harmful overall results: “Staffs in the routine grades will be condemned to longer periods of monotonous work than if the marriage bar had been enforced.” This work was thought to be so demoralizing that changing the rules of the Civil Service in a way that could keep workers there indefinitely would injure the very workings of the state itself: “Any general worsening of promotion prospects among the large bodies of staff engaged on routine work would, by its adverse effect on staff morale, be against the interests not only of the staff themselves but of the public service,” warned the committee set up to reconsider the bar. Because this class of workers was growing by the year, the potential for it to become unruly and discontented seriously worried Civil Service executives.¹²²

Class as well as gender played a role in the marriage bar. An interwar commission that considered the bar from 1929 through 1931 had argued that it should not be removed in the lower job classes—but perhaps in the higher classes.¹²³ This proposal helped the already privileged and continued to lay a burden on the least powerful. It showed how the government’s unwillingness to rethink the valuation of its employees tended to widen divisions and strengthen inequalities even during the passage of reforms designed to lessen inequality. Later, with equal pay, this pattern would be repeated.

As a result, these reforms had disproportionate effects on women who worked with automating equipment in government and industry. Since the very machinery of the government bureaucracy relied on these temporary, low-paid workers, high turnover through forced resignation upon marriage was one way to ensure a constant supply of this indispensable, if undervalued, commodity. The marriage bar created a self-fulfilling prophecy of deskilled and expendable women workers; because women would resign on marrying, they were never given career opportunities that would encourage them to stay. “And as the time and expense spent on the training of those resigning is lost to the service, women as a whole are, for this reason amongst others, less valuable than men as a whole,” reported the Whitley Council Committee—the body tasked with protecting workers’ rights in the Civil Service.¹²⁴ That the body designed to protect worker rights viewed women as less valuable than men highlighted women’s precarious position. Little wonder, then, that women were passed over for better training and more responsible work, confined instead to gender-segregated, partially automated work that was held in low esteem regardless of its actual content or complexity.

The 1944 to 1946 committee set up to consider equal pay pointed out that this system of widespread gendered job segregation flew in the face of the ideal of job “aggregation” that was supposed to allow Civil Service employees to navigate the career rungs of the Service’s meritocracy. Aggregation was supposed to allow civil servants to move from one post to another across departments, and to be promoted in a consistent fashion even after changing jobs. For women, aggregation barely applied: “While in some Departments women are employed on the same duties and work side by side with the men, in others they are employed in separate branches and have separate avenues of advancement. ... The latter system, if rigidly observed, amounts to the reservation of certain posts to men and women respectively,” the Royal Commission on Equal Pay noted.¹²⁵

Immediately after the war’s end, the government surveyed a variety of organizations—including city and local government offices, the nationalized industries, and private companies—and ultimately decided to lift the Civil Service marriage bar in 1946, though equal pay remained unaffected. They had failed to remove the bar before World War II, pointing to divided opinion within labor unions, but after the war the main clerical union vociferously supported the measure because its membership was now mostly women.¹²⁶ The British Broadcasting Corporation (BBC), London County Council, and Boots Drug Company also removed their marriage bars after the war, stating that no major disadvantage could be found in employing married women. Other national governments like the USA and USSR opposed the concept of a legal marriage bar, though their cultures encouraged it to different extents. Sweden, Denmark, and Finland also stood opposed. Meanwhile, Commonwealth countries like Canada, Australia, and South Africa returned to their prewar ban on married women workers.¹²⁷ Many large British employers, like the Bank of England, made a provision for married women whose husbands had died during the war but reinstituted their overall policy of not employing married women, as did the four main railway companies, the Cadbury and Rowntree Co., and most local government offices.

The disaster the government feared from the marriage bar’s repeal would turn out to be a phantom; a decade after repeal most women still continued to leave upon marrying. A cartoon from the 1950s (figure 1.4) poked fun at the depressing reason. In it a pretty, young woman, chained to her typewriter desk, declared her intention to marry a slovenly ne’er-do-well in order to escape: “You can’t be worse than my job,” she says, running toward him.¹²⁸ Stuck for the most part in positions that had little career potential, most women in government found the prospect of working after marriage understandably unattractive.

Figure 1.4 Cartoon from Mavis Tate, MP, Equal Pay Campaign Pamphlet, “Equal Work Deserves Equal Pay,” 1954.

The government’s expectation that women would resign on marriage aligned with a general perception of women workers as an elastic and incidental labor force who conveniently inhabited a holding pattern doing unpaid work in the home when not needed in the paid labor force. However, the idea that most women only required a wage until marriage masked a more complex situation. Seen as temporary workers, women were usually given less responsible jobs that were not likely to lead to careers. One woman trade unionist groused that “it was not for any high moral principles that [industry] was getting rid of their married women ... they wished to employ cheap labor,” and employers often “seized on any opportunity to worsen the conditions of employment” for married women, on the assumption that a married woman already had the needed level of income through her husband.¹²⁹ As expendable labor, women’s jobs often followed automating tools, and this in turn created organizational methods that kept their wages depressed.

In the context of growing postindustrial workforces, increasingly automated offices, and higher numbers of working women, resistance to removing the marriage bar had as much to do with economics as culture. Government ministers rightly conjectured that agitation for equal pay and promotion would increase once women’s working careers were lengthened by the removal of the bar, and feared this would worsen labor problems. The new expenses of maternity leave, and pensions for greater numbers of women working after marriage, might significantly raise labor costs.¹³⁰ Lower-order Civil Service executives feared that the lack of turnover and promotion outlets in these feminized job grades could ignite a labor powder keg. It was an untenable situation.

As a result, shortly after the marriage bar’s removal and on the recommendation of a government commission, the Treasury department, with its broad oversight over staff, orchestrated a change in the very structure of the Civil Service itself. Less than two years after the death of the marriage bar, the “machine grades” were created, and women’s economic dependence on husbands was replaced by their economic dependence on machines. The latter dependence, just like the former, would ensure that women’s wages remained depressed. This new class of machine-operating work would interact catastrophically with the campaign for equal pay: It would institutionalize the connection between low-paid women workers and computing for decades to come.

The Expansion of Computerization as a Tool of the State

In 1945, Alan Turing presented a proposal for an “Automatic Computing Engine” to the National Physical Laboratory. His paper outlined a stored-program electronic computer that would revolutionize data processing by vastly increasing the speed at which input and output could be run through the computing unit. Through eliminating “all that is done by the normal human operator” and transferring these functions to the machine, this system could vastly reduce what Turing called “the human brake.”¹³¹

Turing’s concept of a human brake on computers’ efficiency and his research on how to circumvent it were early steps in the study of human–computer interaction. Through programming that could be stored inside the machine’s memory, he reduced the human brake at the level of hardware, but the brake persisted elsewhere. As large computerized systems made up of machines, workers, and organizational entities grew, opportunities for “braking” multiplied—not everything could be subsumed by hardware.¹³² The difficulty of dealing with the relationships between machines and humans inherent in computing systems could not be designed away. This would become a major stumbling block as the computer moved out of solely scientific and military operations and into a wide variety of commercial and governmental workplaces, its sphere of influence continually growing. The importance of the human brake would increase as faster digital, programmable computers designed with business and administrative applications in mind began to replace mechanical and electromechanical office-automation machinery.¹³³

Defined by its wartime origins, electronic computing was arguably Britain’s most critical twentieth-century project. Hidden, wartime computer operators had a growing analogue in peacetime: the “broad base” of the Civil Service’s data-processing establishment, which rested on feminized and largely ignored classes of machine workers. As a tool in times of stability and peace, not just in times of war, computerization’s history is largely a narrative of the expanding reach and power of the data-intensive state.¹³⁴ The peacetime organization and deployment of electronic computers within the government’s Civil Service and the nationalized industries would soon become a microcosm of the nation’s attempts to grapple with large-scale technological change.¹³⁵ Although the groundbreaking deployment of electronic machines by the British government remained secret for decades after the war, the labor patterns that allowed those systems to function persisted. The same gendered labor assumptions that helped Bletchley Park maximize its wartime codebreaking also strongly impacted how the national government organized automation and large-scale information-processing projects after the war. Although the war helped women gain greater opportunities and a springboard to argue for greater work equality, these benefits soon collided with the structural discrimination of the government’s large-scale data projects.

As British offices and institutions computerized, the interaction of labor and management with new technology became essential to the maintenance and expansion of the government’s power. Successive governments recognized harnessing the power of these early computing systems, which were often untested and unwieldy, as a crucial component of national progress. However, early electronic computers did not naturally lead to a streamlining of public sector bureaucracies. Human interaction with automated systems formed a critical facet of the history of early computing because the British government modeled the organization of new information systems on older office-automating machinery, which was in turn designed around a specific kind of deskilled, feminized labor force. Computer deployment in government would become unhelpfully fixated on the removal of the “human brake.” Despite Britain’s early lead in the field of computing, this labor landscape undercut the nation’s “revolutionary” technological project.

Although women faced a postwar labor market in which higher-level technology jobs were very difficult to get, they also entered a labor landscape where their femininity marked them as machine workers. In this way, neither the war nor its end was a watershed moment for the gendering of computing labor. What the war had done, however, was put a spotlight on the dangers of having a gendered class structure that produced particular kinds of workers for different jobs. This gendered class structure resulted in a scramble to quickly retrain large numbers of women for the war effort. Once the labor shortages of wartime ended, however, these inequalities once again became a secondary concern. Until, that is, operators’ labor once again became recognized as a critical part of the computing systems on which the nation relied.

1 War Machines: Women’s Computing Work and the Underpinnings of the Data-Driven State, 1930–1946