When Computers Were Human

by David Alan Grier

  Florence Weldon proved to be a greater help to her husband than Galton’s committee. “The committee did not possess a mathematician to put on the break,” claimed Shaw’s friend Karl Pearson, “and Weldon attempted too much in too short a time.” As W. F. Raphael Weldon began to publish his results, he was met by the same kind of hostility that had greeted the calculations of Halley and Clairaut. W. F. Raphael Weldon’s work was far from perfect, and his mathematical formulas did not always demonstrate the properties he had hoped to illustrate; but the response to his work was not in proportion to its flaws. “The very notion that the Darwinian theory might, after all, be capable of statistical demonstration seemed to excite all sorts and conditions of men to hostility,” observed Pearson. W. F. Raphael Weldon worked with the committee through the mid-1890s and then took a position at Oxford University.23

  The methods of organized statistical calculation, especially the calculation of correlation values, developed in the laboratory of Karl Pearson at the University of London. Pearson was a professor of mathematics, but he had broad interests that ranged from history and politics to religious faith and the relations between the sexes.24 Born Carl Pearson, he was the son of a London attorney and the product of a traditional Cambridge mathematics education, including the third-place finish on the Tripos. As a young man, he had a crisis of faith that caused him to abandon conventional Christianity and embrace socialism. After two years of study in Germany he adopted the German spelling of his first name, Karl.25 Pearson was a radical but not a bomb thrower. George Bernard Shaw described such people as Pearson as “unconventional in a conventional way.” “[He] was in many ways poorly socialized,” observed his biographer, Ted Porter, “a thoroughly original character who, while drawing deeply and repeatedly from the cultural resources of his time, rejected many of the conventions of his class and profession.”26 Before turning his attention to statistical theory, he wrote books on the philosophy of science and organized a selection of his friends into a Men and Women’s Club. According to a historian of the club, “discussions ranged from sexual relations in Periclean Athens to the position of Buddhist nuns, to more contemporary discussions of the organization and regulation of sexuality, particularly in relation to marriage, prostitution, and friendship.” Pearson’s presentations were highly intellectual, often laced with Darwinian ideas, and were occasionally beyond the grasp of other club members.27

  Pearson’s influence over the practical issues of computation began in 1895, when the Royal Society added Pearson to the Committee for Conducting Statistical Inquiries into the Measurable Characteristics of Plants and Animals. Pearson was not particularly impressed by the organization of the group or its method of operation. He later recalled that “Weldon’s work was hampered by the committee” and suggested that the members had neither the inclination nor the ability to help him.28 Pearson’s contribution to the group was a mathematical formula for correlation, a formula that turned correlation analyses from a lengthy graphical procedure requiring a certain judgment to a straightforward equation. This formula required the computers to summarize the data in five quantities. In Galton’s example, two of the quantities were computed from the children’s heights, two more came from the mid-parent heights, and the last was calculated from products of the two sets of data. A final computation of four multiplications, three subtractions, one square root extraction, and a long division produced the correlation value.29

  The formula for correlation became one of the first mathematical tools for a small computing group that Pearson formed at the University of London in an organization he would call the “Biometrics Laboratory.” “All the work of computing undertaken in my Department,” Pearson explained, “[was] entirely done by volunteer workers,” a group of computers that included students, friends, relatives, and his wife.30 Though Pearson dominated the group, he liked to think that they were all collaborating as equals. The first large project of this group began in the summer of 1899. At “Hampden Farm House in the Chilterns,” he reported, “we had at our disposal a considerable strip of garden covered with Shirley poppies.” The poppies, with their distinctive seed pods, provided the basic material for a study of inheritance. Pearson recruited fifteen friends to help with the research, a collection of eight men and seven women that included the Weldons. Several of this group had been members of the Men and Women’s Club. Pearson treated this project as a socialistic endeavor, an effort in which all contributed equally. Though he was directing the work, he took his turn with the more mundane tasks, such as tending the plants, measuring specimens, and harvesting the seed.31

  Pearson’s socialism did not prevent the summer from having an informal elegance, the air of a comfortable English country house during the last summer of Queen Victoria’s reign. The group would meet on Fridays, at what Pearson called the “biometric teas,” a term that invoked men with high collars, women in long dresses, and an attentive servant pouring tea and offering cucumber sandwiches. In the leisure of a long summer afternoon, they would review their progress, discuss new ideas, and make plans for the weekend. Pearson reported that “Saturday and Sunday … were given to calculation and reducing weekly work.”32 In this case, the term “reduction” meant not the processing of astronomical data but the calculation of the five basic terms for correlation analyses. The results of this experiment were published in a paper without an author, though a footnote acknowledged that Pearson had drafted the report. The paper listed the contributions of all sixteen workers, including the women. It also invited interested parties to join the research. “To any of our readers willing to assist in further observations on ‘first flower,’ the Editors will most gladly send seed of pedigree poppies with suggestions for further work.”33

  17. Karl Pearson with Brunsviga calculator

  The cooperative spirit of Hampden Farm would quickly dissipate, but at least a few of the workers would have a long relationship with Karl Pearson and the Biometrics Lab. One of the summer workers, Alice Lee (1859–1939), became a student of Pearson’s. She held a bachelor’s degree from Bedford College, a women’s college at the University of London. She supported herself by teaching mathematics and physics at the college, occasionally covering Greek and Latin courses whenever there was a need. To supplement her income, she lived in the college rooms and oversaw the students. This assignment gave her little time to herself, but it paid for her room and board.34

  Initially, Lee worked as a volunteer in the Biometrics Lab. Her relationship to Pearson was awkward at best. At times she was his student and at times a staff member. Occasionally, he thought of her as a peer but never as an equal. Twice she declined to be listed as the coauthor of a paper with Pearson. She wrote Pearson, “I have done nothing but the Arithmetic, and I suppose a machine could do most of that.”35 She may have been modest about her accomplishments or scrupulously honest in her dealings with Pearson, but she may also have suspected that indiscriminate credit might hurt her reputation as a scholar. If her name appeared on a paper for simply doing arithmetic, then other scientists could conclude that she was nothing more than a computer. She was attempting to establish a reputation in the field of craniometry, the measurement of skulls. The task is hard enough if the subject is dead but considerably more difficult if the object of measurement remains alive. She developed a statistical model that estimated the cranial volume of living subjects from external skull measurements. When she presented the results of her study, some faculty claimed that the work was a simple elaboration of Pearson’s ideas. Pearson intervened in the evaluation of her work, defended the merits of her approach before his colleagues, and argued that he had no claim to her results. Based upon his presentation, the faculty reconsidered the case and awarded the degree.36

  Pearson’s treatment of Alice Lee and his other computers vacillated between radical ideas and common stereotypes. He could defend the contributions of women to Francis Galton, writing that “their work is equal at the very least to that of the men. They are women who
in many cases have taken higher academic honours than the men and are intellectually their peers.”37 Yet in private he could express doubts about the quality of their work and was not always able to treat women as equals. He once complained of Lee, “I am here, as on other occasions, apt to be vexed by her want of power of expression.” Rather than blame her education or even her upbringing, he found the cause in her gender. “On the whole I think it is characteristic of most women’s work.”38

  Through 1903, Pearson had no regular source of funds to support the Biometrics Laboratory. “I live on students fees practically,” he wrote to a colleague, “and any step which leaves my department under incomplete supervision tends to impair its efficiency [and] affect my income.”39 That year, he received a grant of £500 from a civic organization that had been established by the businessmen that manufactured and traded cloth, an organization known as the Worshipful Company of Drapers.40 The funds brought some measure of order to the laboratory and a salary to Alice Lee. She received £90 a year, about half the salary given to an assistant professor. For this money, she worked three days a week, arriving at 9:30 AM and leaving at 5:00 PM. Pearson allowed her half an hour for lunch, which was presumably taken at her desk or, when the weather allowed, in the courtyard of the school. Her duties included reducing data, computing correlation coefficients, creating bar charts—charts that Pearson was now calling “histograms”—and calculating a new kind of statistic, which Pearson had denoted χ2 (usually pronounced chi-squared). The χ2 statistic promised to be just as important as the correlation statistic. It allowed researchers to test scientific theories in a formal, mathematical way.41

  In addition to her computational work, Lee did “all the hundred and one things that need doing here.” She acted “more or less” as the laboratory secretary, excerpted books in the library, formed indexes, and organized catalogs of data. By the standards of office work, the job was not well paid, a fact that slightly embarrassed Pearson. A young female typist or stenographer was usually able to feed a mother or a child on an office worker’s salary. He confessed, “It is a post well suited to a woman living with her family in London and keen on scientific work.”42 Even with all of this clerical work, Lee continued to pursue her own research projects, which led to four papers published in her own name and contributions to twenty-six others.

  The Draper grant allowed Pearson to hire the sisters Cave-Browne-Cave as part-time computers. Beatrice and Frances Cave-Browne-Cave were the daughters of a senior civil servant and graduates of Girton College. They met both Pearson and Francis Galton at a “reading party,” a meeting that was probably a public discussion of natural inheritance or statistical methods.43 The elder, Beatrice Cave-Browne-Cave (1874–1947), taught mathematics in a girls’ school located in south London. She worked as a part-time employee of the Biometrics Lab, doing calculations in her home. Before the Draper grant gave her a small stipend, she did Pearson’s calculations without compensation. The only reward she received was to be listed as a joint author on two papers, one published jointly with Pearson, the other as a collaborative effort of five authors.44

  The younger sister, Frances Cave-Browne-Cave (1876–1965), remained at Girton College as a teacher. She had been the top student in her class and had tied the Fifth Wrangler in the Cambridge Tripos exam.45 Like Alice Lee, she had a research program of her own. She was performing a correlation analysis of weather data that had been collected from the east and west sides of the Atlantic Ocean. Pearson guided the work and outlined the basic mathematics but let Cave-Browne-Cave work at her own pace. She was a popular teacher at Girton and enjoyed socializing with the young women at the college. She devoted many of her evenings to her students, talking with them and helping them with their studies.46 Only on those rare occasions when she was left alone in her room did she find time for her research. “The magnitude of the computations,” she recorded, “almost precluded the idea that any individual worker or workers can hope to complete such a task within a reasonable period.”47 The project required her to calculate hundreds of correlations with Pearson’s formula. Even though she worked on her own, she was able to complete two substantial projects that demonstrated patterns in the weather as it moved across the ocean.48

  Even with the Draper grant, Pearson’s largest projects retained the collaborative aspects of the Hampden Farm experiment. In 1903, he oversaw a large study of child development, “a cooperative investigation extending over a number of years, and depending upon a body of collaborators.” The project collected physical measurements and character assessments from 4,000 children and their parents in order to establish evidence of the inheritance of what Pearson called “moral qualities,” attributes that we would now identify as aspects of intelligence or personality. Both sisters Cave-Browne-Cave were among the six collaborators who worked on the project. Unlike the Hampden Farm experiment, this project seemed firmly in the control of Pearson. He was the one setting the scale of the project and posing the questions to answer. His collaborators gathered the data by measuring and observing the children. Beatrice Cave-Browne-Cave collected data from her high school students. Only a few of the assistants, including both Frances and Beatrice Cave-Browne-Cave, processed the data, created tables, and computed the correlations.49

  In 1904, the Worshipful Company of Drapers pledged an annual grant to the Biometrics Laboratory, giving Pearson a measure of financial security.50 With this money, the laboratory slowly lost its collaborative feel and acquired the more conventional feel of a university office. The lab workers split into three distinct strata: professors, students, and staff. Most of the staff worked as computers, and most, though not all, of the computers were women. These women occupied an isolated corner of academic life. They tended to interact either with Pearson or with other women. Lee taught women at Bedford; Frances Cave-Browne-Cave was a professor to women at Girton; her sister Beatrice worked at a girls’ high school. Their experience was not that different from the life of other female computers. At Harvard, the observatory computing staff had been augmented by women who measured photographs and conducted research of their own. Most of these women were aware that their positions were quite different from those of the men. “I have to see to all the changes of household linen, etc. and gather together the family wash,” wrote a member of the Harvard Observatory staff. “Alas! How matter of fact and different from the Sunday morning duties of other officers of the University.”51

  The increasing stratification of labor in the 1890s made the computing laboratories vulnerable to the labor troubles of the age, though none of them faced anything as severe as the strike at the Homestead steel mill in Pennsylvania or the riots in Chicago’s Haymarket Square. The most dramatic incident occurred when the Greenwich Observatory was bombed by a French anarchist. The explosion shook the building, though it left the structure undamaged and the computing staff untouched.52 Investigators looked for a motivation for the explosion among the observatory employees and found only minor complaints. Among the human computers, the most pressing concern was the policy, established by George Airy, of dismissing any computer once he or she reached the age of twenty-three. “The mathematics and kindred subjects which have been acquired [at the observatory] become absolutely useless,” complained one former computer, “as men are not wanted in the market unless having had a good business experience. So they swell the already overcrowded unskilled labour multitude. This rotten system seems to be maintained by the Government solely to save money.”53 Investigators found it difficult to connect such sentiments to the group that prepared the bomb. They eventually concluded that the explosion was detonated accidentally and that the observatory was probably not the target. The writer Joseph Conrad fictionalized the incident in his novel The Secret Agent. He suggested that the bomber was targeting the institutions of capitalism and the British government, not the observatory itself.54 In the story, the bomb was thrown “into pure mathematics,” and it had “all the shocking senselessness of gratuitous blasphemy.�
�55

  Most of the labor problems among computers were the ordinary conflicts between labor and management and hence were more irritating than dramatic. At the American Nautical Almanac, the “gentlemen of liberal education” had been replaced long ago by computers drawn from the civil service rolls. Computers had to pass a special exam to gain a place on the almanac staff. This exam was difficult, and it regularly failed to identify enough qualified applicants to fill the vacancies in the computing room.56 The director of the almanac office was Simon Newcomb (1835–1909), who had begun his career as a computer under the direction of Charles Henry Davis. Through his work at the almanac, he had achieved a certain measure of fame and was probably the best-known American scientist of his day. He had a conservative nature and had little sympathy for the complaints of workers, labor movements, and strikers. When he was asked his opinion on the growing use of machines in manufacture and their impact upon workers, he replied that “the laboring man is earning higher wages now than he did … before the introduction of labor saving machinery.” He watched the labor unrest of the 1890s with a sense of alarm and concern. He placed the blame for such discontent directly on the workers. “Too many men will not do more than they are compelled to do,” he said; “they do not work cheerfully and become malcontents ready to destroy.”57

  For the most part, Newcomb’s computers were German immigrants from Foggy Bottom, the working-class neighborhood of Washington that served as a home to the almanac office. His most promising computer was a Swiss-German named John Meier. Meier “was the most perfect example of a mathematical machine that I ever had at my command,” Newcomb reported. Meier was hardworking and skilled at arithmetic. Newcomb also observed that, “Happily for his peace of mind, he was totally devoid of worldly ambition.” Meier lived the turbulent life of urban working classes and regularly needed Newcomb’s assistance “as an arbitrator of family dissensions.”58 Meier suffered from an illness that he called “nervosity.” Newcomb gave no name to the disease, though he clearly believed it to be alcoholism. Meier, who had been able to limit the problems caused by his “nervosity,” began to lose control of his life when his wife left him. His children, a boy and a girl, proved more than he could handle alone. His son, testy and combative, showed that he was more than ready to pick fights with his father. The daughter, seventeen years old, had no one who could discipline her and was often found “in company with young men.”59 Newcomb avoided intervening in the failing marriage, but he advised Meier’s son, counseled the daughter, and sought support from the family’s pastor, the minister of the neighborhood German church. After several months, Newcomb tired of the demands upon his time and concluded that Meier simply was not capable of working for the almanac. He relieved Meier from service and requested the return of all books belonging to the almanac office. Using the popular notions of inheritance to justify his actions, he wrote that Meier “illustrates the maxim that ‘blood will tell’” and then added, “which I fear is as true in scientific work as in any other field of human activity.”60