When Computers Were Human

by David Alan Grier

  This was a strange and mysterious war zone but I suppose that it was quite well run and grim compared to other wars. …

  Ernest Hemingway, A Farewell to Arms (1929)

  BY THE SUMMER OF 1918, the whole European war seemed to be packed into the city of Washington, D.C., an area barely ten miles on a side. The army tested ordnance in the city’s northwest quadrant while the navy built guns in the southeast. Residents rented any available room to the new war workers who arrived each day by train. Downtown, entire government bureaus were squeezed into offices that once had held three employees, or two, or even just one. Outside, construction crews labored to create new buildings on the one large, unoccupied piece of land in the center of the city: the parklands of the Washington mall. Watching all of this activity from his office, the director of the army’s Department of Ballistics concluded that the government needed nothing so much as time, “time to build manufacturing capacity on a grand scale without the hampering necessity for immediate production; time to secure the best in design; time to attain quality in enormous output to come later as opposed to early quantity of indifferent design.”1 For those who worked in Washington, the war was a problem of production more than it was a problem of military strategy. The offices in the city had to clothe, equip, and feed the members of the American Expeditionary Force. They had to transport every soldier to the front, provide the forces with weapons and medical care, and return each veteran to the United States. The tools of their war were typewriters, forms in triplicate, adding machines, accounting sheets, and statistics. For many a government manager, the most powerful machine was the punched card tabulator. “Calculators … went to war,” wrote historian James Cortada, “but the dramatic examples of data processing at work were punched card gear.”2

  The tabulator of 1918 was substantially more sophisticated than the simple counting machine of 1890. Following the success of his first census, Herman Hollerith had created a tabulator that could add numbers and accumulate sums. He devised a method for representing numbers as holes in a card. A number was punched into the card one digit at a time, with the location of each hole representing the value of the digit. According to historian Emerson Pugh, Hollerith created this tabulator in order to sell his machines to the railroad companies. “His primary target was the New York Central, which had become the second largest railroad system in the United States.” He promoted the tabulators as a tool for processing waybills, the paperwork that followed the movement of freight.3 He also found a ready market for the new adding tabulators within the United States Census Bureau. In 1900 and 1910, the Census Office had used the tabulators to gather agricultural statistics. Such a process was beyond the ability of the original tabulators, as it required the machines to sum the number of cultivated acres in each township, the total cattle in each county, and the bushels of grain produced in each state.4

  The 1910 census had been the last for Herman Hollerith. In 1911, Hollerith sold his company to a syndicate of investors, who merged the company with two other firms to create the Computing, Tabulating and Recording Company, or CTR. The president of this new company was Thomas J. Watson (1874–1956), who had made his reputation as a salesman for the National Cash Register Company. Watson saw a great future in punched cards and worked to find new markets for the equipment in manufacturing, banking, retail, and government.5 When the United States went to war in 1917, he was ready to provide tabulating gear to any government division that needed to prepare statistics, to direct military operations, or to manage war production.6 With CTR tabulators, the United States government created nearly a dozen new computing offices during the war, each office with a production capacity that rivaled the abilities of the Census Bureau. All of these offices were associated with temporary agencies that managed the wartime economy, and all of them processed substantial amounts of statistical data. The largest of these offices was part of the War Industries Board, the agency that oversaw American manufacturing. This board, under the direction of financier Bernard Baruch, worked to provide the American Expeditionary Force with all the equipment it needed while, at the same time, ensuring that the civilian economy had an adequate supply of goods and services.

  The agriculture production board, known as the Food Administration, was directed by Herbert Hoover (1874–1964). Hoover was then known as a successful mining engineer and the chair of a committee that had successfully brought food relief to Belgium. Hoover faced one of the most complex managerial tasks of the war. The country had hundreds of thousands of food producers, ranging from the kitchen gardens of housewives to the massive plants of the meatpackers. As Hoover started assembling his staff in one of the new buildings on the mall, he recognized that the United States government had only a vague understanding of how producers and processors delivered food to the American market and what resources might be redirected to the war effort without causing undue hardship at home. The government’s principal research office on food distribution, the Bureau of Markets, had been created only four years before by the United States Department of Agriculture. Hoover moved quickly to form a comprehensive statistics office and appointed as its director Raymond Pearl (1879–1940).

  Raymond Pearl came to Washington from the Agricultural Experiment Station for the state of Maine. He was one of the many scientists, like Oswald Veblen and Forest Ray Moulton, who were eager to bring their expertise to the conflict. He had spent nearly ten years as the chief scientist of the experiment station, which was a common name for an experimental farm. At the station, he had overseen tests of new farming methods, fertilizers, equipment, plant stocks, and animal breeds. Pearl had arrived at the station with a traditional degree in zoology and a year of study with Karl Pearson at the Biometrics Laboratory. He had traveled to London in the fall of 1905 and had learned the statistical methods that Pearson and his colleagues were using to study the problems of natural inheritance. His first publication after leaving Pearson looked like nothing more than a freshwater version of the research done by W. F. Weldon and Florence Tebb Weldon in the 1890s. Instead of looking at crabs, Pearl had studied variations in crayfish. At the Maine Agricultural Experiment Station, he took the statistical methods that he had learned from Pearson and put them to work on the concrete problems of improving farm output.

  Arriving at the Food Administration in July 1917, Pearl had to put experimentation aside in order to assemble a staff capable of processing a large flow of data. In many ways, his computing office resembled the Smithsonian project that had summarized weather statistics in the nineteenth century. At the center of his organization was a punched card processing room with a staff of several dozen machine operators. Surrounding this facility were several smaller offices, each charged with preparing statistical reports on a specific aspect of the agricultural economy. One office handled sugar production, another food storage, and a third retail prices. These smaller divisions collected data from a staff of volunteers, prepared cards for the tabulating machines, and edited the final reports. The division that dealt with retail prices was typical of these offices. It began operation in September 1917 with “a few clerks and stenographers,” according to Pearl. It “grew and expanded until … [it] consisted of about thirty five people.”7 Each week, it received data from its volunteer staff, “a selected number of reliable people in every town and city of 3,000 or more inhabitants.” The volunteers, who were almost exclusively women, in contrast to the male computing staff, were given a package of forms and instructions to collect the prices of specific items. “In the majority of cases, the prices reported were those taken by the reporter from her own sales slips,” wrote Pearl. The completed forms were sent to Washington once a week. Seven hundred of these forms arrived in the first week of October, the first week of tabulating data. This number grew to 1,000 by the first of December. By the end of the winter, the retail price division was processing 2,000 forms a week.8

  In general, the numbers produced by Pearl’s tabulators, clerks, and computers were used by Hoover an
d the senior leaders of the Food Administration to describe the state of the agricultural economy. They were rarely used to explore the operation of agricultural markets and suggest policies that might increase production. If the leadership of the Food Administration had been astronomers in the days of Edmund Halley, they would have been content to review the known observations of the comet of 1682 and accept the notion that it returned to view every 75 years, more or less. They would have been unmoved to compute the comet’s orbit or use it as a test of Isaac Newton’s theory of gravitation. The notion that the sprawling agricultural economy could be described with differential equations or probed with statistical calculations was not widely accepted in 1917–18. Long after the war was over, the Washington Post ran an editorial that mocked attempts to study the economy mathematically as “Hog astronomy” that required computations with “Hogarithms.”9 While some of this mockery came from the general resistance to scientific analysis, a resistance that paralleled the objections made by Jonathan Swift to the calculations of comet orbits, part was political. Many involved in business did not want any intervention in the market that might restrict their ability to make a profit. Hoover preferred to appeal to the “voluntary effort of the people” in order to achieve his goals and placed his trust in “the spirit of self-denial and self-sacrifice in this country.”10 The records of the Food Administration are filled with posters and advertisements that encouraged the recycling of fat, the preparation of meatless meals, the production of more pork, and other ways of making the best use of the nation’s food.11

  Hoover’s approach to agricultural management often conflicted with ideas proposed by specific parts of his administration. Each of the major food products had a board or commission that reviewed the reports from Pearl’s statistical office and recommended policies. There were commissions for all of the major agricultural products, including beef, sugar, milk, and wheat. These commissions were more familiar with the production of their specific commodities and were often more willing than Hoover to intervene in the markets. It was within these commissions that the power of punched card tabulation met the mathematics of scientific calculation.

  The Swine Commission, which was described as seven “expert swine men from all over the United States,” has left us the clearest record of its experiments with advanced mathematics.12 Well before the start of the war, the production of pork was already handled in a highly systematic way. Most of the country’s pork came from an area known as the “Corn Belt.” Roughly this area was centered on Chicago and stretched from Iowa on the west to Indiana on the east. The northern border of the Corn Belt could be found in Minnesota and Wisconsin, while the southern edge was in Missouri. “Corn-belt agriculture was integrated into national markets almost from the time the Midwest was settled,” wrote agricultural historian Mark Friedberger.13 Pork production began with corn, which Midwest farmers grew as feed for their hogs. When the hogs were grown and fattened, they were loaded onto trains and shipped to slaughterhouses in Chicago. Rail lines led directly into the buildings where the hogs were killed and butchered on long, moving “disassembly lines.” Once it was dressed and packed, the meat was loaded onto refrigerated rail cars and shipped to the urban centers of the East. The process of turning corn into bacon and hams and chops was described by Upton Sinclair, the great novelist of meatpacking, as “porkmaking by machinery, porkmaking by applied mathematics.”14

  At the Swine Commission, the applied mathematics of pork-making was developed by a young newspaper editor from Iowa named Henry A. Wallace (1888–1965). Wallace served the commission as an unofficial staff member. He had come to Washington as an assistant to his father, Harry C. Wallace (1866–1924), one of the seven “expert swine men.” The Wallace family paper, named Wallace’s Practical Farmer, had a strong circulation in the central Corn Belt. It offered readers snippets of gossip, reports of agricultural news, market prices, advertisements, Bible lessons, recipes, advice, and general guidance for the farmer, all under the motto “Good Farming, Clear Thinking, Right Living.”15 Through the paper, the elder Wallace spoke with a respected voice on agricultural, social, and political issues. He was a member of the Republican Party, though he was part of the party’s progressive, pro-agriculture wing.16

  The younger Wallace, Henry A., had a great faith in numbers, in the ability of data to reveal the problems of society and suggest a governmental policy that would improve the life of ordinary people. It was a faith that was scientific and yet was deeply rooted in his family’s religious heritage. Throughout his life, he would return to a biblical story of famine. “Behold, there come seven years of great plenty throughout all the land of Egypt: And there shall arise after them seven years of famine; and all the plenty shall be forgotten in the land of Egypt; and the famine shall consume the land.”17 The hero of this story was a young man named Joseph, one of the sons of Israel. Through his spiritual insight, Joseph foresaw the famine and showed the Egyptians how to prepare for the years of dearth. When the famine arrived, Egypt had food enough and to spare, “corn as the sand of the sea, very much, until he left numbering; for it was without number.”18

  At the start of the war, America had no want of food but the lot of the farmer was precarious. In general, farm income had been declining since the end of the Civil War with only a few periods of increase. “Who will be like Joseph?” Henry A. Wallace asked in an article he wrote for Wallace’s Practical Farmer.19 Though he never rejected spiritual solutions, he looked to numbers and statistics to provide the guide for the nation’s farmers. After completing a degree in agriculture at Iowa State College, he had started a systematic study of mathematics and statistics. He had learned calculus with the assistance of a professor at Drake University in Des Moines. When he was ready to learn the basic concepts of mathematical statistics, he acquired a book that had been written by a member of Karl Pearson’s Galton Laboratory and studied the work himself.20

  At the Swine Commission, Henry A. Wallace found the opportunity to apply his mathematical knowledge in a computation that described the relation between the corn and hog markets. His father had argued that “hogs are simply condensed corn” and that farmers would produce more hogs only if they could make more profit by selling pork than by selling corn.21 If this were true, then the government could increase hog production by guaranteeing a higher price for hog flesh than for the equivalent amount of corn. The younger Wallace invented a value he called the “corn-hog ratio,” which was the price of a bushel of corn divided by the price of one hundred pounds of hog flesh, and began the calculations to justify his father’s theory. The work was not as difficult as the 1758 computations of Halley’s comet, but they would have required a substantial computing staff if Wallace had not been able to use the machine-tabulated reports of Raymond Pearl’s statistical office. He computed corn-hog ratios by hand from these reports and concluded that hog production would increase if the guaranteed price for one hundred pounds of hog flesh exceeded the price of thirteen bushels of corn.22

  When presented with the calculations of Henry A. Wallace, Herbert Hoover was unable to repeat the blessing that the biblical pharaoh had bestowed upon Joseph: “Forasmuch as God hath shewed thee all this, there is none so discreet and wise as thou art: Thou shalt be over my house, and according unto thy word shall all my people be ruled.”23 Hoover viewed the proposal as an unacceptable form of government intervention in the marketplace and had no intention of implementing it. If anything, he wanted to remind the Wallaces of the last line of the pharaoh’s blessing: “in the throne will I be greater than thou.”24 This disagreement marred the last two months of the war for the Swine Commission. Hoover remained ever obdurate, and Harry C. Wallace attempted to force the issue by bringing public pressure to bear on the man who was credited with saving the people of Belgium from starvation. Wallace wrote articles in his family farm paper and recruited the assistance of other papers in the Corn Belt. At times, the language of this campaign was abusive, suggesting that, perhaps, the c
orn-hog ratio did not have the same heavenly source as Joseph’s plan for Egypt or that the biblical pharaoh was more accepting of new ideas than Woodrow Wilson’s food administrator. As the Allied troops began their final campaign against Germany, Harry C. Wallace accused Hoover of acting with “such chicanery and deceit as an experienced businessman knows how to use in case of emergency.”25 Unmoved, Hoover held his position.26

  Had there been more time, had the United States been engaged in the war for a longer period, the Food Administration might easily have been forced to undertake more mathematical analyses of the agricultural economy and find policies that were acceptable to all, but as the country entered the last month of the conflict, discipline at all of the major computing organizations began to falter. At Aberdeen and at the Experimental Ballistics Office in Washington, the computers started growing restless. The final campaign of the war coincided with the start of the academic year, and a few computers left the proving ground in order to return to their universities. Among those that remained, a gentle discontent began to grow. Forest Ray Moulton, who had become slightly bothered by the military discipline, quipped, “It has sometimes been a little irritating for men of national and international reputations as scientists to be compelled to show their photographs before they were permitted to stand in line at a cheap cafeteria.”27 Others tired of the military discipline that invaded their mathematical idyll. One computer wrote that Aberdeen was a “queer sort of environment, where office rank, army rank and academic rank all played a role, and a lieutenant might address a private under him as ‘Doctor’ or take orders from a sergeant.”28

  Part of the restlessness came from the realization that only a small part of the Aberdeen computations were going to reach Europe. Early in the war, American ordnance officers had decided to delay the deployment of American guns in order to release space on the transit ships for troops and other supplies.29 The gunners of the American Expeditionary Force used French and British weapons. Though a few range tables for these weapons were prepared in the United States, the majority of the calculations had been done at the Galton Laboratory in England or at Gâvre in France. The only prominent weapons to make the Atlantic crossing were large naval guns that were mounted on railroad cars. The guns were deployed in September and quickly moved toward the retreating German army. In an action that could be interpreted as a fit of pique or as a show of Allied dominance, one railroad gun battery used range tables prepared at Aberdeen to place the last shot of the war. The battery crew fired the shot at 10:57 AM on November 11. It flew for nearly two and a half minutes, striking its target seconds before the armistice time of 11:00 AM.30