When Computers Were Human

by David Alan Grier

  Among the tens of thousands of WPA projects, there were about 2,400 that supported scientific research. Future Nobel laureates Glenn Seaborg and Luis Alvarez at the University of California are on the list of those who received WPA assistance.5 WPA funds paid the salaries of laboratory workers in much the same way that the National Youth Administration had provided funds for student assistants. Most WPA-sponsored science involved large statistical studies of social or economic problems.6 Malcolm Morrow’s division of the WPA, the Central Statistical Office, oversaw most of these projects. This office suggested possible studies, reviewed proposals from universities and state agencies, and granted funds that would pay for workers to conduct surveys, collect information, tabulate data, and analyze the results. The Iowa State Statistical Laboratory was a major beneficiary of the Central Statistical Office. It received WPA money to summarize harvests, measure farm income, and analyze the extent of rural poverty.7

  Sometime early in the operation of the WPA, perhaps in the summer of 1935 or the winter of 1936, the staff of the Central Statistical Office became engaged in a discussion of mathematical tables. As they were starting to oversee a large number of statistical studies, they thought that it might be useful to provide a standard set of mathematical tables to each of the projects and to each of the regional WPA offices. If all of the statistical projects had the same tables of logarithms and probability functions, then the Central Statistical Office could compare the results of the different studies with a little more confidence. This idea produced no immediate action, as the office had more pressing things to do, but it was revived in the fall of 1937 when the nation entered a new period of economic hardship. After three years of tentative growth, the economy had begun a contraction that was quickly labeled the “Roosevelt Recession.” Faced with growing unemployment and greater labor unrest, the WPA began looking for new projects in order to expand the number of relief jobs. With winter approaching, they needed to identify activities that could operate during the cold months.8 The construction jobs, which constituted 75 percent of the WPA projects, were curtailed between late fall and early spring. As WPA officials evaluated different ideas, they recognized that a computational project could employ a large office staff and assigned Malcolm Morrow the task of organizing such a group.

  The mechanics of starting a new project were fairly simple. Morrow needed to find office space, acquire the necessary furniture, and notify the local WPA office that he could accept a staff of qualified workers. The local office would send him as many people as he needed from those who had applied for relief jobs. The difficult part of Morrow’s assignment was the task of finding a sponsor, a scientific organization that would accept the project, oversee the computations, and provide the appropriate mathematical expertise. He started looking for a sponsor at the institution that considered itself to be at the center of American scientific research, the National Academy of Sciences. The academy building stood just a few blocks from the mammoth headquarters of the WPA, so one day in late October, Malcolm Morrow left his desk early and called upon Albert Barrows, the permanent secretary of the National Academy, and asked for assistance.

  Morrow began his presentation slowly, stating ideas that were familiar to many of the scientists who worked with the academy. He reported that the statisticians on the WPA staff felt that “there was a very large amount of work, which can be done to improve the mathematical basis on which a great deal of other scientific work depends.” He noted that there were substantial errors “in a number of the standard mathematical tables” and that the “tables already in use ought to be greatly extended,” as more scientists were employing them in their research. Reaching his central point, he told Barrows that the WPA was planning to build a large computing center in New York City. The center would prepare mathematical tables for use “not only by mathematicians and astronomers, but also by surveyors, engineers, chemists, physicists, biometricians, statisticians, etc.” If they were successful, it would be the biggest computing organization in history, a group that might “employ a thousand people, as well as experts to advise.” All of the computers, of course, would be drawn from New York’s unemployment rolls, but he assured Barrows that there would be calculating machines and a “training school to initiate employees.” The new center would be an expensive operation, Morrow conceded, but New York City had $2.3 million available for relief. He implied that a large part of this money “could be devoted to this project.”9

  After describing a computing facility nearly as fantastic as Lewis Fry Richardson’s weather computing room, Morrow retreated to a more practical realm and admitted that the project would begin on a small scale, perhaps “50 employees under the supervision of technical directors and office managers.” Coming to the key question of the meeting, he asked Barrows if the National Academy of Sciences would be willing to appoint an advisory committee to the computing center, a group that would identify projects for the computers and help prepare computing plans.10

  In general, the members of the National Academy of Sciences belonged to a class that distrusted Franklin Roosevelt and disliked the idea of government-sponsored work relief. They joked that the initials “WPA” stood for “We Poke Along” and claimed that the public works projects were nothing more than an attempt to buy votes for the Democratic Party. This view of work relief was so pervasive among the secure classes of American society that twenty years after the end of the Great Depression, the novelist Harper Lee could damn a character by claiming, “He was the only man I ever heard of who was fired from the WPA for laziness.”11 If Barrows held such impressions about work relief or the WPA, he was able to put them aside as he summarized the meeting for academy president Frank Lillie (1870–1947), a University of Chicago zoologist. “Mr. Morrow (a man I should judge of 40 or 45 years of age) impressed me as competent to discuss this project,” he wrote.12

  Lillie responded to Morrow’s visit by sending a note to the chief statistician of the WPA. “I wish to assure you that we shall be most happy to render any service within our power,” he wrote; “the National Academy of Sciences in its part would be glad to appoint a committee of mathematicians to advise on specific undertakings.”13 Lillie named no names, but he and Barrows had already discussed possible candidates for such a committee, including Oswald Veblen, Vannevar Bush, Harold T. Davis, and James Glover.14

  When Malcolm Morrow ended his meeting at the National Academy of Sciences, he had told Barrows that there was “an element of urgency in organizing the matter because of the impending election of mayor in New York City,” and he had asked that “no intimation of consideration of the project be permitted to be made publicly or find its way into the channels of the press.”15 The WPA computing office, tentatively called the Project for the Re-computation of Mathematical Tables, was not an ordinary scientific project that was supported by a university or funded by industry. It was going to be a federal project in a city where Democrats and Republicans fought over federal funds and where the outcomes of such battles could have national repercussions. Morrow was probably concerned about protecting his project, getting the computing office established before some city agency attempted to capture its budget for another purpose. The sitting mayor of New York City, Fiorello La Guardia, was an ally of Franklin Roosevelt and willing to have projects managed from Washington.16 His opponent, a member of the city’s Tammany Hall political organization, was not as sympathetic to the Roosevelt administration and might be able to redirect the project funds to some organization with no qualifications other than ties to Tammany Hall.17

  After his meeting at the National Academy of Sciences, Morrow traveled to New York City, hoping to persuade one of the city’s universities to sponsor the project. Columbia University was the most likely choice, as they had a strong mathematics department and housed the offices of the American Mathematical Society. However, its faculty already operated the Columbia Computing Bureau with its tabulating machines. They showed no interest in sponsoring a large group of
human computers. Morrow then approached New York University and proposed that the school provide space for the computers and mathematicians to manage the calculations. Though university officials were interested in the project, they felt that they could not provide the necessary funds and declined the offer.18

  As the date of the mayoral election approached, Morrow had no sponsor for the project, no site to house it, and no plans for its operations. At the National Academy of Sciences, which had received no report from Morrow after his initial visit, secretary Barrows voiced a growing skepticism about the idea. “I still have the feeling,” he complained, “that this is one of those matters which they would not be able to complete if started.”19 Morrow found his sponsor three days before the November election. After finding no help in New York, he returned to Washington and convinced the director of the National Bureau of Standards, the physicist Lyman Briggs (1874–1963), to manage the new computing organization.20

  Briggs had both the technical background and the political experience for such an assignment. As a physicist, he knew how hard it could be to prepare a useful, error-free table. From his years with the Bureau of Standards, he understood that government support for science was unstable in the best of times and hostile in the worst. Since its founding in 1901, the bureau had been repeatedly attacked by congressional critics, who had argued that the government should not be involved in research and should not set standards for private industry. In 1933, Congress cut the budget of the National Bureau of Standards in half. “It was a bitter experience for us,” Briggs recorded. “More than one third of our staff was dropped on a month’s notice.”21 Attempting to sustain the bureau, he discovered that he could obtain relief grants to replace some of his lost budget. Money from the Civil Works Administration and the Public Works Administration, predecessors of the WPA, was used to clean and repair bureau buildings.22 “Several of the abler mechanics and technicians of the Bureau,” wrote one historian, “let go earlier, found their way into [maintenance work] and tarried there until they could be restored to the Bureau payroll.”23

  In agreeing to sponsor the WPA computing project, Briggs extracted concessions that New York University was unable to get. He would manage the project, recruit senior personnel, and provide scientific expertise, but he would provide no money for the group. The WPA would fund the entire budget for the project. Furthermore, Malcolm Morrow, who had identified himself as the project leader, was relegated to administrative issues, such as finding a suitable office, handling WPA paperwork, and communicating with the New York City relief agencies. This agreement was consummated so quickly that Morrow apparently failed to tell Lyman Briggs of his contacts with the National Academy of Sciences, for the politically astute Briggs made no effort to contact Albert Barrows or Frank Lillie. With no information from either Briggs or the WPA, the senior members of the academy naturally concluded that the relief agency had been unable to organize the computing project. “Since the day Mr. Morrow was in here saying we were to get a letter the next day, I have wondered what has happened to him,” commented one member of the Academy staff. “He seems to have disappeared off the horizon.”24

  It would have been best for the Mathematical Tables Project if Lyman Briggs had been able to recruit an established computer to be its leader. If the project had been led by Oswald Veblen or even H. T. Davis, it would have started operations with strong connections to the scientific world. However, the idea of leading a work relief project appealed to few scientists, and Lyman Briggs had to turn to the ranks of the underemployed, the scientists who were not able to find a position at a top college or university. The path that led him to choose Arnold Lowan (1898–1962) as project director is no longer well marked. It may have involved Oswald Veblen, probably passed through Columbia University, and almost certainly involved thermodynamic calculations, the computations of heating and cooling.

  In the fall of 1937, Arnold Lowan was holding two part-time teaching positions. By day, he taught physics and mathematics at Yeshiva College in Manhattan. At night, he taught the same subjects at Brooklyn College.25 He had immigrated to the United States in 1924, fleeing anti-Semitic pogroms in his home country of Romania and rushing to reach New York before a new law restricted immigration from Eastern Europe.26 He was a desirable immigrant, a chemical engineer, but he had waited four years before he found professional employment, a period he would never describe on any of his resumes or reminiscences of the time. He likely spent the time among the immigrant Jews of Brooklyn or Manhattan’s Lower East Side, learning English and becoming acclimated to the United States. If he was employed during this time, he was likely doing manual labor or other work that he considered subprofessional. The first job he was willing to identify was one he took in 1928 with a utility contractor, working as what he called a combustion engineer. He described his job in grand terms, claiming that he was researching the “thermodynamics of gaseous hydrocarbons,” while he was actually working with coal gas furnaces and boilers.27

  30. Arnold Lowan, director of the Mathematical Tables Project

  The salary he gained from tending furnaces by day allowed him to study physics by night. He took a master’s degree from New York University and then transferred to Columbia University in order to earn a doctorate. “He certainly has great industry and perseverance,” remarked a Columbia physicist, and “in the main [he] follows out his own ideas.” At a time when the brightest students were looking at problems that would lead to the subjects of relativity and quantum physics, Lowan turned to conventional thermodynamics, the study of heat. Though he rarely asked the Columbia faculty for guidance, he did become acquainted with one of the veterans of the Aberdeen Proving Grounds, a physicist who had served as Oswald Veblen’s assistant during the First World War. This connection gave Lowan a brief entry into the senior ranks of American scientists, a postgraduate fellowship at the newly formed Institute for Advanced Study in Princeton, where Veblen was on the faculty.28

  At the institute, Lowan seems to have kept to himself and rarely interacted with the other researchers. He spent his year developing his ideas on thermodynamics and completing a large calculation that described the cooling of the earth. In all, he did the research for ten scientific papers, each of which involved extensive numerical work. He may have been overawed by scientists like Veblen and Albert Einstein, or he may have had difficulty communicating with others, or he may simply have lacked the background to deal with research problems in the new fields of physics.29 Judging from his later letters, he clearly spent some time with Einstein, Veblen, and the Hungarian mathematician John von Neumann (1903–1957), but not enough to be their friend or even their familiar. In 1934, when Lowan’s fellowship ended, they were unable or unwilling to help him find a job at a research university. With the lingering depression, budgets were tight; Lowan could not find any full-time job in physics. Instead, he took his two part-time teaching positions.

  There are three plausible routes that might have led Lyman Briggs to Arnold Lowan. The shortest begins with Oswald Veblen, the great computer of the First World War, who would have recalled Lowan’s computational work at the Institute for Advanced Study. The second path is more complex, but it was suggested by one of the WPA workers. This path begins with one of Lyman Briggs’s assistants, a geologist who had prepared mathematical tables for the Smithsonian. For this path to work, the geologist would have to have read Lowan’s articles on the cooling of the earth and appreciated the extensive computations. The final path leads through Columbia University. Briggs may have simply asked the Columbia faculty or the American Mathematical Society if they knew of anyone qualified to run a computing organization.30 No matter how Briggs was introduced to Lowan, he could not have found a leader better suited for the Mathematical Tables Project. Lowan was ambitious to make a reputation as a scientist and was not repulsed by a work relief job.

  Lowan joined the Mathematical Tables Project in late November 1937 and was told to begin operations in the first months of the ne
w year. The project office would be one floor of an industrial building on the west side of Manhattan, not far from Times Square. Arnold Lowan found a used desk to serve as his work area, placed it in a corner of the empty space, and started to make his plans. He would be the executive of the group, the administrator who would deal with budgets, personnel, correspondence with the public, and any communications from the Bureau of Standards. His first problem was to find more mathematical talent, a technical director to analyze the computations and prepare the detailed computing plans. Almost immediately, he hired two graduate students from Brooklyn College, but they acted as personal assistants, not mathematical leaders. His technical director needed to hold a doctorate and should know something about the operations of a large office. He probably had no idea where to search for such an individual and must have been surprised to find one in his night class at Brooklyn College.31

  That fall, Lowan was teaching a course on the subject of relativity, an elementary presentation designed for individuals who were trying to expand their horizons. Most of the students came from day jobs in Manhattan. Sitting near the back was a short woman who wore the kind of business dress that could be washed in a bathroom sink and dried overnight. On most nights, she seemed a little tired and not especially engaged in the subject. When it was cold or raining, she did not attend the lecture at all. Lowan paid little attention to her until he began to grade the homework papers. Her mathematical reasoning was the work of a professional. It identified all the assumptions, moved through each step of the analysis, and presented the solutions in a clear manner. As the student used mathematical concepts that had not been presented in class, Lowan suspected that she was receiving outside assistance. He traveled to and from campus on the same bus line as this student, so one night after class, he sat next to her and tried to start a conversation. At first, the student was reluctant to speak, as if she was unsure of Lowan’s motives. After some coaxing, she began to relax and tell her story.