When Computers Were Human

by David Alan Grier

  Veblen implemented this plan on all of the Aberdeen ranges, though some of the facilities lacked proper equipment. On one of the antiaircraft fire ranges, before the computers acquired range finders, they resorted to techniques that might have been borrowed from Francis Galton or a scientist of Laputa. They would fire the shells vertically into the night sky and photograph the bursts against the background of the fixed stars. After the photographs were developed, the computers would measure the distance between the shell burst and the stars. With this data, they would use the tables of the nautical almanac to compute the altitude of the burst.55

  Aberdeen had no residences for women and hence had no female computers. The only women who worked at the base were a pair of secretaries, who were forbidden to spend the night at the proving ground.56 The army began hiring female computers only when it opened an office of experimental ballistics in Washington, D.C. This office was directed by Major Forest Ray Moulton (1872–1952), who was a professor of astronomy at the University of Chicago before he accepted a reserve commission. Moulton took the job of preparing ballistics materials for ordnance officers. In the spring of 1918, he was given an office in a temporary building on the Washington Mall and told to hire a staff. As other officers had already discovered, Moulton found it difficult to hire enough men, so he offered positions to women.

  The young women of 1918 could not attend the special training summer camps, volunteer for a Canadian regiment, or even fire a 12-inch gun at the Aberdeen Proving Ground. Though they were generally enthusiastic about the war, their feelings were checked by the roles that they were offered in the conflict. The first year of the war was also the year of women’s suffrage, the year that a corps of committed women moved to Washington, D.C., in order to win the right to vote. While the men were preparing to fight in France, women were picketing the White House, lobbying the members of Congress, and marching up and down Pennsylvania Avenue. One congressional aide recalled seeing “cultured, intellectual women arrested and dragged off to prison because of their method of giving publicity to what they believed to be the truth.”57

  The suffrage movement surrounded the world of Elizabeth Webb Wilson (1896–1980). Wilson was the daughter of a Washington physician, a flaxen-haired, dimple-cheeked member of the capital’s wealthy classes. She studied mathematics at George Washington University, a school just a few blocks west of the White House. Her daily trolley ride took her past the lobbyists, the pickets, and the marches. Like herself, many of the suffrage leaders were the daughters of physicians. As far as we know, she took no part in the effort that ultimately caused President Wilson to endorse the suffrage amendment, Congress to approve the measure, and the states, one by one, to give their consent. Yet, in the spring of 1918, she took her own small stand for women’s equality. When she heard that the federal government would employ women in war offices, she applied for a job that would “release a man for the front.” It was the “patriotic thing to do,” she recalled. But when she was offered a position, she refused it on the ground that it was “insufficiently mathematical.” She had been the top mathematics student in her graduating class, the first woman to win the school’s mathematics prize. Though her college peers had judged her a quiet and timid woman, she stood her ground and stated that she would only take a war job where her mathematical talents “could be utilized to the fullest.”58 The personnel office offered her a second position, which she also declined, and then a third. In all, she rejected nine jobs before accepting an appointment as Moulton’s chief computer.

  21. Elizabeth Webb Wilson in the spring of 1917

  Elizabeth Wilson’s persistence was a small victory for feminism. Washington had already established itself as a city of opportunity for single women. The novelist John Dos Passos described contemporary stenography offices in the capital, where “the typewriters would trill and jingle and all the girls’ fingers would go like mad typing briefs, manuscripts of undelivered speeches by lobbyists, occasionally overflow from a newspaperman or a scientist.”59 A woman could earn seventeen or twenty dollars a week, enough to pay the rent on her own apartment, send some money to her family, and have enough left to occasionally purchase clothes from one of the finer department stores. The Washington office of experimental ballistics employed both men and women and put them in situations where they had to work together. The computing staff had sixteen workers and was split equally between female and male. Like Wilson, the other seven women were the offspring of prosperous homes and graduates of coeducational schools with strong mathematics programs. Two of the women had studied at the University of Chicago, two at Brown University, and one each at Cornell University, Northwestern University, and Columbia University.60

  The staff of the Washington ballistics office stood midway between the mathematicians and engineers of Aberdeen and the officers in the army’s artillery corps. Using data from the Aberdeen ranges, they prepared tables and documents for those who would actually command the guns in battle. The work was often sensitive or political in nature, requiring Moulton to negotiate among different branches of the military to establish standard operating procedures. “As a consequence of the various interests involved,” he wrote about one problem, the work “had to be taken up somewhat formally.”61 In this environment, the mathematicians worked closely with the computers to prepare material that was both accurate and appropriate for the situation. During Wilson’s first weeks in the office, she and Moulton had to prepare a range table for a French gun firing American ammunition. Wilson demonstrated “both personal mastery of the technical operations involved,” wrote one of the mathematicians in the office, “and skill in supervising and checking the work of others.”62

  Reflecting on the experience, Moulton saw an unusual camaraderie between the mathematicians and the computers. “The unfailing courtesy and the evidence of mutual helpfulness which were manifested in numerous ways,” he wrote, “were inspired not alone by military customs and the proprieties of the situation, but much more by sincere mutual respect and personal regard.” Wilson, the only member of the office staff from Washington, acquired the role of chief computer and social leader. She hosted a dinner for the office staff at her parents’ home and a party at her father’s club.63

  When the ballistics computers recalled their service in the First World War, they would remember the summer of 1918. “It would be difficult to gather in any way an equal number of individuals who would have more in common and whose relations would have been more harmonious,” wrote Moulton.64 The number of firings at Aberdeen increased each day, and the data flowed from the gun mounts to the proving ground’s computing building and from there to the experimental ballistics office in Washington. Veblen spent much of the summer traveling in search of new computers, but whenever possible, he would catch an early train back to Aberdeen so that he could join the artillery crews on the range. He would fire the guns until the dimming light made further observations impossible.65 He recruited mathematical talent from universities, from industry, and even from the offices of the Encyclopedia Americana. “The demand was immediate,” remembered one computer, so “[I] terminated my [job]. I took the next train to New York, where I changed for Aberdeen.”66 In less than two months, Veblen added twenty-three graduate students or new PhDs to his staff, bringing the total number of computers at Aberdeen to thirty. Twice that summer, the army engineers had to double the size of the computing building.67

  During those months, the computers gathered range data for naval guns mounted on railroad cars, tested new designs of streamlined shells, and uncovered a major design flaw in existing shells, a problem that Moulton judged to be “so great that the guns were of little value.”68 Yet the problem that most interested them was a major revision of Siacci’s ballistics theory. “Upon entering the army,” wrote Moulton, “a hasty examination of the classical ballistic methods showed … that they were wholly inadequate for current demands.” He argued that Siacci’s analysis “contained defects of reasoning, some quite erron
eous conclusions, and the results were arrived at by singularly awkward methods.”69 His criticisms were unduly harsh, as the First World War armies were using the theory for problems that Siacci had not foreseen. Siacci had not planned for the problems of high-altitude fire, antiaircraft guns, and long-range artillery, nor had he anticipated that an army might have at its disposal a staff of forty-six human computers.

  By all accounts, Siacci’s theory continued to work fairly well for short- and intermediate-range artillery, but events in the first years of the war had shown that it failed dramatically in long-range artillery and high-angle mortar attacks, in addition to its shortcomings for antiaircraft fire, bombing, and fire from aircraft. While the staff would be able to correct some of these deficiencies, Moulton concluded that he should develop a new, comprehensive ballistics theory that could be used to analyze any circumstance. The work engaged the computing staffs in both Aberdeen and Washington and challenged them to use the most sophisticated mathematical concepts at their disposal. Moulton created the central outline of the theory. Other mathematicians handled specific problems with the theory, such as adjustments for altitude, the spinning of the projectile, or the rotation of the earth. Elizabeth Webb Wilson, who helped prepare the tables for the theory, recalled discovering that “the Germans had the advantage because the earth turned toward the east, therefore as they were shooting toward the west, their bullets carried further into the allies’ lines.”70

  The computers were mathematicians, not ordnance engineers, and were most interested in the mathematics of Moulton’s ballistics theory. One computer, ignoring the advantages to the gunnery crew, recalled how the work “made a brilliant use of the new theory of functionals,”71 a concept that was then on the frontier of mathematical research. Isolated from friends and family, separated by an ocean from the dangers of war, the computing staff at Aberdeen lived the life of extended adolescence. They hiked through the countryside and conducted unauthorized, and probably unscientific, experiments with TNT and smokeless powder. From the collected scraps and rubble of ordnance experiments, they invented surreal versions of checkers and chess. At night they would gather in the computing shack, play cards, and do the things that young men do when they are at war, though perhaps they were unique in calculating their winnings and loses on adding machines which by day computed the trajectories of shells. After the cards were dealt, the conversations would wane as the players computed the probabilities that their opponents held winning hands. When they spoke, they talked not of lost opportunities or of distant family members, but of the mathematics that they loved and the theorems that they would prove.72 One computer wrote that the experience “furnished a certain equivalent to that cloistered but enthusiastic intellectual life which I had previously experienced at the English Cambridge.”73

  The trenches in France had their own intellectual life, at least for the English troops. “The efficiency of the postal service made books as common at the front as parcels from Fortnum and Mason’s,” wrote critic Paul Fussell, “and the prevailing boredom of the static tactical situation … assured that they were read as in no other war.”74 For the scientists at the front, there were opportunities to observe and speculate. The English meteorologist Lewis Fry Richardson (1881–1953) found that many of his days were uninterrupted by combat or even by rumors of combat. The battles most often occurred at sunrise or sunset, when one side or the other was shielded by the glare from the low-lying sun. He served as an ambulance driver and had little responsibility beyond the work of caring for his vehicle. During the quiet hours, he worked at his science, theorizing about the movement of the winds, the distribution of humidity, the impact of the light from the sun. “My office was a heap of hay in a cold rest billet,” he recalled. His approach to the problem was not a statistical method, like that used by the U.S. Weather Service in the 1870s, but a differential equation model that had much in common with the mathematics of astronomy and ballistics.75

  Richardson described his analysis of the weather as “a scheme of weather prediction, which resembles the process by which the Nautical Almanac is produced.”76 He derived a series of differential equations that described how the weather changed moment by moment. These equations tracked seven basic properties of the atmosphere: its movement (in three dimensions), density, pressure, humidity, and temperature. The computing plan for these equations divided the globe into “a special pattern like that of a chessboard,” a grid of longitude and latitude lines that marked 2,000 points where the weather would be computed in increments of three hours. “It took me the best part of six weeks to draw up the computing forms,” he recorded, and when he attempted the calculations, he discovered that he required an equal amount of time to calculate a single advance of the weather at one of the points. His duties at the front had prevented him from fully concentrating on the arithmetic, and at one point, he had misplaced his manuscript. “During the battle of Champagne in April 1917,” he recalled, “the working copy was sent to the rear, where it became lost, to be re-discovered some months later under a heap of coal.”77 When he finished the calculations, he concluded that “with practice, the work of an average computer might go, perhaps, ten times faster.” From this one exercise, his little respite from the war, he concluded that it would require 32 computers to keep pace with the weather at one grid point, 32 computers to complete a single three-hour prediction in exactly three hours. As there were 2,000 points in his scheme, he would require total of 64,000 human computers to track the weather for the entire globe.78

  “After so much hard reasoning,” Richardson asked, “may one play with a fantasy?” He had not come to France in search of glory. He was a Quaker and a conscientious objector to war. Rather than serve in the military, he had chosen to drive an ambulance. His fantasy was a world where the soldiers massed on the Western Front were put to work reproducing the earth’s weather in numbers. He would take 64,000 soldiers from the front lines and make them computers in a giant spherical computing room, a room that would have been larger than any sports stadium of Richardson’s day. The internal “walls of this chamber are painted to form a map of the globe,” he wrote. The North Pole would be on the ceiling, while Antarctica would be marked on the floor. England would be found three-quarters of the way toward the top, its shape nearly hidden by one of the many balconies that ringed the inside of the room. Upon these balconies, Richardson imagined, “a myriad computers are at work upon the weather of the part of the map where each sits.” He suggested that the computers might work on large sheets of paper and then display their results on “numerous little night signs” so that others could read them.79

  The calculations would be directed by a senior computer, a scientist who had worked through every step of the mathematics. This computer would stand on the top of a tall and slender column that rose from the floor like the column for Admiral Nelson in Trafalgar Square or a tall skyscraper in a city of lesser office buildings. “Surrounded by several assistants and messengers,” the senior computer would “maintain a uniform speed of progress in all parts of the globe.” Richardson suggested that this computer “is like the conductor of an orchestra, in which the instruments are slide-rules and calculating machines.” In his scheme, the conductor’s baton was replaced by a pair of colored spotlights. The computer would turn “a beam of rosy light upon any region that is running ahead of the rest, and a beam of blue light upon those who are behindhand.” Outside of the computing room, the computer oversaw a scientific compound, which included a radio station to transmit the weather predictions, a secure storeroom to hold old computing sheets, and a building that held “all the usual financial, correspondence and administrative offices.” Next to the sphere was a training room and a research lab, as “there is much experimenting on a small scale before any change is made in the complex routine of the computing theatre.” Richardson’s weather compound ended at the border of Arcadia, the land where numbers stood for things of nature, rather than the flight of artillery shells, the
output of a factory, or the commerce of men. “Outside are playing fields, houses, mountains and lakes,” wrote Richardson at the end of his fantasy, “for it was thought that those who compute the weather should breathe of it freely.”80

  The fantasy of a giant computing laboratory was, perhaps, not so unrealistic to those who served on the battlefields or carried the wounded to safety. During his three years with the ambulance corps, Richardson saw hundreds of miles of trenches that were filled with thousands and thousands of soldiers. All he had imagined was the simple act of winding those trenches into a ball and using their occupants for the peaceful end of computing the weather. However, in 1918, weather did not represent the same kind of threat as the German army, so that no government had any interest in building a laboratory for 64,000 computers, 6,400 computers, or even 640 computers. Only one facility approached Richardson’s vision of a computing compound with houses and lakes and trees. It employed but forty-two computers and was located in a Maryland town named Aberdeen.

  CHAPTER TEN

  War Production