When Computers Were Human

by David Alan Grier

  The supporters of the “remonstrance” were uncomfortably close to Davis and his friends. Their leader, Ingersoll Bowditch (1806–1889), was the director of a Boston insurance firm, a friend of Benjamin Peirce, and a financial supporter of the Harvard Observatory. Their objections to a meridian at New Orleans were entirely economic. Bowditch believed that his business would be damaged by the change in meridians. He was a partner in a firm that republished the British Nautical Almanac in the United States. The objections of the other merchants and insurers were more speculative. They were concerned that the ports of Boston and New York would decline if New Orleans had a meridian passing through it. New Orleans already possessed a substantial advantage over the Atlantic ports, as it had a navigable waterway, the Mississippi River, with access to the vast center of the continent. With the meridian, reasoned Bowditch and his supporters, New Orleans would become a destination for ships needing to adjust their chronometers. These ships would divert trade to the south, as none of the captains would want to travel empty.

  Davis attempted to counter the claims of Bowditch, but he had lost the battle almost from the beginning. He argued that his proposal was “founded on principles of science” and would contribute “towards improving the safety of navigation, and completing the geography of the seas,” but neither idea swayed his opponents.49 The debate had turned toward economic issues and way from scientific merit. In this field, Bowditch held the greater power. By the spring of 1850, Davis had abandoned his idea for the meridian and accepted a solution that divided the American almanac into two parts. The first part, the American Ephemeris, would be computed relative to a meridian that passed through the Naval Observatory in Washington. This meridian would be used by surveyors as they set the borders of Wyoming, Colorado, Oregon, and the other western states. These lines would fall at integral number of degrees west from the Naval Observatory in Washington, rather than from the Royal Observatory in England. The second part of the almanac, the Nautical Almanac, would be prepared using the Greenwich meridian and be used by the nation’s sailors.

  The double meridian scheme put extra demands upon the almanac computers, as it required them to prepare more tables, but it had little impact upon the structure of the computing staff. Davis did not have to restructure his computers in order to make them more efficient because he had substantial support from both the navy and Congress. The navy had initially allocated $6,000 a year for the almanac, most of which was spent on the salaries of computers. By the spring of 1850, Davis had concluded that this figure was insufficient, especially with the requirement to prepare two sets of tables. He requested and received $12,000 for the second year of operations. This figure also proved to be too little. For his third year of operations, Davis asked for $18,000.50 The navy granted this request, but the increase drew the attention of the U.S. Senate. In May of 1852, Senator John P. Hale (1806–1873) of New Hampshire rose from his desk on the Senate floor and asked the navy to justify its expenditures on the almanac. Though he expressed his concern over the size of the almanac budget, Hale was more interested in a key element of Davis’s plan, the cooperation of naval officers and a civilian computing staff. He might have been less concerned if the computers had been “mere drudges” and could have been drawn from any state of the union. However, he knew that all of Davis’s computers were somehow connected to Benjamin Peirce or Harvard College.51

  Hale demanded that the secretary of the navy “inform the Senate, where, and at what Observatory, the observations and calculations for the ‘Nautical Almanac’ are made.” Like any skilled politician, Hale knew the answer to his question before he strode onto the Senate floor and asked to be recognized. “I think that I am not incorrect,” he informed his colleagues, “when I say that all this expense has been incurred, not at the National Observatory but at the observatory of Cambridge College in Massachusetts.”52 “Cambridge College” was, of course, Harvard. He probably misidentified the school as a way of emphasizing its location. If he expected this charge to stir the Senate to action, he was disappointed. Even those who objected to government support for private institutions sat in their seats. Any discussion of the almanac was ended quickly and decisively by a senator who derided Hale as possessing “an absolute and unappeasable hostility to any connections of science with the Naval Department in any form.”53

  Though Hale’s comments lasted but a few moments, they pushed Charles Henry Davis to respond. His plans would quickly fail if Congress reduced his budget or dictated the kinds of computers that he might hire. He wrote a detailed defense of his organization, addressed to the secretary of the navy but published in the American Journal of Science and Arts and circulated as a pamphlet. He confronted Hale on the narrow point of the attack, explaining that “no Observatory, neither that at Washington nor that at Cambridge, as has been suggested, received any portion whatever of the sum appropriated for the ‘Nautical Almanac.’” This statement was entirely true, but it did not address the point that many of the computers had a connection to Harvard. Davis tried to deflect this issue by stating that the almanac required the “most illustrious genius and the most exalted talents” and that it was not “a work of insignificant value or trifling labor.” He claimed that the new almanac “is considered by American astronomers and mathematicians as a work of consummate utility and of real national importance, resembling in this respect the Nautical Almanac and Astronomical Ephemeris of Great Britain, the Connaissance des Temps of France and the Astronomical Almanac of Prussia.”54

  The almanac remained vulnerable to congressional displeasure, even though the crisis initiated by Senator Hale passed quickly. Through the early 1850s, Davis worked to ensure that the almanac offered no target for an easy attack. “In our work there can be no regular vacation,” he advised a new almanac computer. Perhaps sensing that he had overstated his case, he conceded that computing was not a full-time activity and that there were “opportunities for occasional indulgence.”55 Without such opportunities, he would have found it difficult to retain his computing staff, as all of them had other interests. Benjamin Peirce requested a reduced computing load so that he might concentrate on mathematics. Maria Mitchell traveled to Europe and relied on her father to communicate with the almanac office.56 The most serious threat to almanac operations was the unexpected death of Sears Cook Walker, but Davis was able to recruit Walker’s brother-in-law as a replacement.57

  The first substantive change in the almanac occurred in 1856, when Davis decided that he must put his naval career ahead of his love for astronomical calculation. In September of that year, the secretary of the navy offered him the command of a ship in the Caribbean. Davis described the order as “an alternative.” He could accept the post and leave the almanac staff; “the other choice was to give up all desire for a command and to resign the active service.”58 No matter which choice he made, he had lost control of the almanac, so he accepted the command, gave his office to the senior computer, Joseph Winlock, and departed for the Caribbean.

  One of the almanac computers described Joseph Winlock as “silent as General Grant with the ordinary run of men.” He could be friendly and open among the computing staff, but “he had a way of putting his words into exact official form.”59 He had no interest in changing the operations of the almanac but was content to follow the computing plans of Charles Henry Davis and the advice of the ever-present Benjamin Peirce. His only step toward innovation involved the almanac in a controversy at the newly formed Dudley Observatory in Albany, New York. That observatory had acquired a difference engine, though it was not the engine that had been designed by Charles Babbage and had never been completed. This engine was a smaller machine that had been built by two Swedes, George Scheutz (1785–1873) and his son Edvard (1821–1888). The elder Scheutz had read about the Babbage difference engine in a newspaper and had built a simple model with Edvard. The younger Scheutz had created engineering drawings and found a machine shop that would build the engine. Their design drew on the technolog
y of clocks, while Babbage had borrowed the tools and ideas of steam engines. The Scheutz engine looked like a large music box and could sit on a desk or a dining table. Recognizing that this machine was an improvement of his ideas, Babbage praised the Scheutz design as “highly deserving of a Medal.”60

  11. Scheutz difference engine used by the computers of the Nautical Almanac

  The Scheutz difference engine had reached the United States because of the efforts of Dudley Observatory director Benjamin Gould (1824–1896). Gould was yet another student of Benjamin Peirce. He had discovered the difference engine while traveling in Europe and had been impressed with its potential. He claimed that it might change the nature of observatory staffs by replacing “the toiling brain by mere muscular force.”61 He had no immediate use for the engine and was willing to let the almanac staff experiment with the device. Joseph Winlock thought that it might be able to prepare certain tables for the almanac. He sent two computers on a trip from Boston to Albany with the instructions to prepare an ephemeris for Mars. When the computers arrived at the Dudley Observatory, they found that the machine was inoperable. “The dirt, which had accumulated on the passage [across the Atlantic], and thickened oil, impeded its action greatly.” A mechanic cleaned the machine and returned it to operating condition, allowing the computers to start their work.62 The calculations proved to be more difficult than anyone had anticipated. The computers discovered that the Scheutz machine was fragile and sensitive. It could easily jam in the middle of a calculation, a problem which could require a lengthy repair effort and force the computers to restart their calculations from the beginning.63

  After a month of work, the computers returned to Boston with two intermediate tables but without a complete ephemeris. They had discovered that the difference engine could calculate only small segments of tables. If the orbit of Mars were placed on a clock dial, the machine could compute the arc between noon and eleven o’clock before it needed to be reset. “The strictly algebraical problems for feeding the machine made quite as heavy demands upon time, and thought, and perseverance, as did the problem of regulating its mechanical action,”64 observed Benjamin Gould. In reviewing the tables, Winlock expressed his hope “that the immense labor of astronomical calculations may be materially diminished by the aid of machinery,” but he was disappointed.65 “The result thus far has not been such as to demonstrate to my satisfaction that any considerable portion of the Almanac can be computed more economically by this machine.”66 The failure of this experiment had more impact in Albany than in Washington. Benjamin Gould had become involved in a fight with observatory trustees over several key decisions, including the purchase of the difference engine. “The responsibility of having recommended this machine I willingly accept,” wrote Gould. “If the machine has not multiplied and tabulated [the donor’s] fame to an amount equal to the wishes of [his] most ardent friends,” he added, “it has not been my fault.”67

  Compared to the other issues facing the almanac office, the failure of the difference engine experiment was only a minor problem for Winlock. The almanac computers had fallen behind schedule. The almanac budget had been reduced by Congress. At the start of 1859, Winlock had been forced to release part of his computing staff and stop work on some of the ephemerides.68 The trouble ended only with the return of Charles Henry Davis that summer. Davis reported “that several parts of the work were omitted or postponed; that printing had been arrested or delayed, and the efficiency of the corps of computers diminished.”69 He quickly brought the computers back to their production schedule and convinced Congress to restore the almanac budget. In less than a year, he could report that “the regular work of the office was resumed with more than common activity.”70

  Of course, the fall of 1860 was no common time. The changing political climate could only remind Davis that the scientific activity of the Nautical Almanac Office was held most firmly in the grip of forces beyond the ability of any one person to control. By December, the goal of preparing a high-quality almanac seemed less important than the need to preserve the country’s social structure. Following the election of Abraham Lincoln in November, the country had reached a point of crisis on the question of slavery, and the Southern states were preparing to secede. Southern naval officers, including the director of the Naval Observatory, resigned their commissions and returned home.71 Charles Henry Davis was summoned to Washington and discovered that he would have little time to think about the operations of the almanac. “I have found that [the director of the Coast Survey] has a plan of his own to carry out, which involves my remaining here,” he explained to his family. “[The director] wishes to establish a military commission, or advisory council, to determine military proceedings and operations along the coast.”72

  Through the spring and summer of 1861, Davis tried to manage the almanac computers while simultaneously organizing surveying units for both the navy and army. He spent most of his days at the Coast Survey building. The survey office was located just south of the Capitol. Its top floors had a view of the navy yard, the Washington arsenal, and the Confederate volunteers of Virginia holding positions on the far side of the Potomac River. The sights and sounds and rumors troubled him so much that he complained, “The more I hear, the more I fear for the end.” He poured out his sadness, his anxiety, and his guilt by invoking Shakespeare’s murderous lord: “Like Macbeth, I’m sick at heart (‘Seyton, I say!’).”73

  The words of Macbeth came freely from his pen that summer. If taken literally, Davis cast himself as the title character, the military leader who had been misled by his own ambitions, had killed his king, and had pushed his country into chaos. Increasingly, the almanac director felt that he was needed elsewhere. “My hands are not of much use in working,” he told his family, “but my head might be in directing.”74 In September, he concluded that he needed to be released from the institution that he had built and led for a decade. “Today I give up the ‘Nautical Almanac,’” Davis wrote. “I am very sorry to do it, but I could not retain it.” He would have to trust his hopes and plans for the almanac to others. “Winlock takes my place,” he acknowledged, “and he will be glad to get it.”75 Though Davis was generally not given to ironic comments, he does not seem to be blessing his successor. An era was ending, and the early computers, the “gentlemen of liberal education,” would soon depart, to be replaced with more traditional computers. Benjamin Peirce had already left the almanac. The others would soon follow. Even Maria Mitchell had only a short time remaining. She would shortly be appointed the first professor of astronomy at Vassar College for Women.

  CHAPTER FIVE

  A Carpet for the Computing Room

  “Arcturus” is his other name—

  I’d rather call him “Star.”

  It’s very mean of Science

  To go and interfere!

  Emily Dickinson, Poem (1859)

  “I WAS SURPRISED to find how quickly one could acquire the stolidity of the soldier,” wrote a computer at the Naval Observatory.1 The observatory, from its hill on the northern bank of the Potomac River, had an unobstructed view of Confederate territory. Directly across the water sat the plantation of General Robert E. Lee, abandoned by its owner and occupied by Union troops. The navy insisted that the computers and astronomers, most of them civilians, be trained as an artillery crew. The staff dutifully reported to the ballistics range at the Washington Navy Yard, where they learned how to charge and fire a cannon. They spent a day sending shells into the opposing riverbank before they returned to their telescopes and computing desks. None of them remarked that the arc of a cannon trajectory bore a mathematical similarity to the path of a comet, and, of course, none of them was quite ready to fire a gun in anger.2

  The war permeated Washington, D.C., filling the streets with uniformed troops and the anxious gossip of distance campaigns. Only once was the city the site of an actual battle. That skirmish was fought on the northern border of Washington, a substantial ride from the solitary cannon tha
t had been deployed at the Naval Observatory. Taken as a whole, the Civil War had little direct impact upon the practice of organized computation, even though it reordered American society, revolutionized American government, and profoundly touched the lives of the computers themselves. For the major American computing offices, the rooms found at the Naval Observatory, the Nautical Almanac, and the Coast Survey, the war was a plague that passed over their institutions, leaving them little changed. Individual computers had to make their choice “twixt that darkness and the light,” in the words of a contemporary poem, but the computing floors could operate as they did before the war.3

  Only the computers of the Coast Survey served in battle. The survey’s computing division was actually the oldest computing office in the country. It had been formed in 1843, the year that the survey was reformed and reorganized. The division hired “gentlemen in private life,” as a report described them, “to repeat the important calculations of the survey.”4 Each summer, the surveyors would disperse to the field to create triangulation surveys, a large network of triangles that connected major landmarks and navigation points along the coast. Inevitably, after the trigonometry and longitude calculations were done, the triangles would fail to align perfectly. Some sides would be too short; others would be too long. Angles would not close, and lines would cross at the wrong place. The surveyors would correct these failures by modifying their numbers, taking a small bit from one side, adding a little to an angle, rounding numbers until all the triangles matched. The Coast Survey computers in Washington duplicated the process of survey adjustment and provided a means of evaluating the judgment of the original surveyors. If the two sets of computations were sufficiently close, then the director would accept the original survey. If they did not agree, then the computers would be asked to prepare a third set of adjusted calculations.5