When Computers Were Human (46 page)

LORAN, an acronym for Long-Range navigation, was a joint project of the Massachusetts Institute of Technology and Bell Telephone Laboratories. It used pairs of special radio stations to guide ships and planes. The two stations would be separated by tens or hundreds of miles, yet they would exchange a special synchronized signal, a little blip that flew back and forth from one station to the other like an electronic tennis ball flying through the ether. Navigators would determine their position by timing the arrival of these electronic blips. From one pair of times, they could determine that their plane or ship was located somewhere on a curved path. If they conducted this operation twice, using two different pairs of stations, they would be able to construct two curves that intersected at the exact point of their position.

MIT engineers demonstrated the LORAN system in June 1942. They used a single pair of radio stations, one located on the tip of Long Island and the other housed in an old Coast Guard Station on the Delaware shore. They placed a LORAN receiver on a navy airship, taught the navigator how to perform the special calculations, and released the craft at Atlantic City. The airship meandered up and down the coastline, going nowhere in particular but giving the navigator a chance to do the arithmetic.
⁶⁹
When the craft returned to its base, the scientific staff declared the experiment a success, even though the navigator was able to do only one-half of the calculations. “We are ready to begin the computations necessary for the production of [navigation] charts,” wrote one of the engineers, but he noted that the navy “cannot now take on the routine but voluminous calculations required.” The Bureau of Navigation, which included the Nautical Almanac Office, did not have enough staff to prepare the tables, and there was no other large computing office within the service. Through Phil Morse, the LORAN engineers learned about the
Mathematical Tables Project. “This group would gladly undertake the calculations,” commented one engineer, “but they already have a number of other jobs progressing.”
⁷⁰

Arnold Lowan was pleased to put aside other jobs for the LORAN work. He assigned the navigation calculations to a small team of computers led by Milton Abramowitz (1913–1958), the longest-serving staff member at the Mathematical Tables Project. Lowan had recruited Abramowitz in 1937, when he was a young graduate student at Brooklyn College. When Blanch had come for her initial inspection of the Mathematical Tables Project office, Abramowitz had been working alone at an old table in the big, empty computing room. In the first years of operation, he was the staff sergeant, the floor leader of the group. As he gained experience and completed his graduate studies, he advanced through the project ranks. He led the special computing group, became a key member of the planning committee, and, during the year of army map grid calculations, took command of the project's punched card equipment.

The MIT engineers estimated that “the services of about a dozen of the better computers would be satisfactory” to prepare the tables, but that was not enough.
⁷¹
When Abramowitz prepared a sample table, he would regularly appear at Lowan's desk, inquire after the director's health, and politely request an additional computer. Lowan, who spoke with a heavy Romanian accent, would complain that he did not have enough computers to fulfill important requests from the army and navy, but in the end he would give Abramowitz the additional worker. By the early fall, Lowan had lost so many computers to LORAN, or to other projects, that he was no longer able to take new assignments. Looking for a way to strengthen his computing staff, he requested that ten of his computers be allowed to work forty hours per week instead of the thirty-two hours mandated by WPA regulations.
⁷²
The New York City WPA office denied the petition, stating that the “Emergency Relief act is intended to furnish temporary work for needy people. It is supposed to encourage workers to gain private employment.”
⁷³

With a shrinking staff and no dispensation to work a full forty hours, the Mathematical Tables Project struggled ahead as best it could. The members of the planning committee accepted some of the extra work, spending their evenings on the computing floor, finishing the calculations, going over results, trying to get the most from their staff. Their efforts were complicated when Lowan received an urgent letter from Phil Morse that raised the issue of security. “I hope,” he wrote, “you have arrangements whereby you are certain that the results of the calculations do not leak out to the outside world without your control.”
⁷⁴
Lowan assured him that human computers had access only to intermediate calculations
and that the senior staff had control of the final results. The reason for Morse's concern soon appeared. In early November, the Mathematical Tables Project completed the sample navigation table for the LORAN project after devoting 5,000 hours of computation to the task.
⁷⁵
The table was reviewed by MIT engineers, who judged the result “very satisfactory” and recommended that the project be given a contract to prepare all of the tables. The navy acknowledged the recommendation but after considering the issues involved, concluded that they would look for some other group to do the calculations. “This decision was based on the security aspects of our work,” wrote one engineer, “and was made by Naval Operations.”
⁷⁶

On December 3, less than two weeks after the LORAN decision, President Franklin Roosevelt announced that the WPA would be terminated and all the projects liquidated. Under the termination plan, the Mathematical Tables Project would operate through March 1943 in order to consolidate its accomplishments and end its activity gracefully. For Arnold Lowan, the news was a blow to the chest, a painful way to end four years of dedicated work. The phrase used by Roosevelt to describe the closure, “honorably discharged,” stung with irony, for Lowan wanted nothing more than to contribute his bit to military research. He was not ready to concede that his computing lab would have to die the death of a relief project, so one more time he returned home to Brooklyn, placed a sheet of paper on his table, and wrote to Philip Morse, “I would welcome any suggestions you may have.”
⁷⁷

When the other members of the Mathematical Tables Project reflected upon the difficult days in late 1942, they liked to believe that L. J. Comrie had saved their jobs. According to Ida Rhodes, L. J. Comrie was listening to the radio on the night of December 3 and heard the news that the WPA would be terminated. “Not bothering to affix his wooden leg, he hopped to the nearest telegraph office and sent a ‘hot wire' to President Roosevelt, stating that under no circumstances must the Math Tables Project be allowed to perish—adding some choice language about boondoggling in other projects.” Upon receiving this telegram, as Rhodes told the story, Roosevelt was supposed to have ordered an investigation. A government inspector, a young woman in a uniform, was immediately dispatched to New York. She called upon the project, reviewed its accomplishments, and recommended that the National Defense Research Council take responsibility for the group. Her opinion carried weight in the highest offices of government, and so the project was saved.
⁷⁸

There is some evidence to support Rhodes's story. Comrie did hear a radio broadcast on December 3 and, a few days later, sent a telegram to President Roosevelt. The message was free of “choice language about
boondoogling” but said simply: “British scientists engaged on war work hope you will provide for continued activity of New York Work Projects Administration Mathematical Tables Project.”
⁷⁹
There is no comment on the artificial leg and no record that the telegram made it from Roosevelt to the office of the WPA. There was too large a gap between the responsibilities of the president and the little project that Lowan oversaw. Furthermore, Comrie had no standing with the government of the United States. He had only limited affiliation with American scientific institutions and commanded no American votes. Even if the telegram had reached the WPA office and been entered into its communication log, the program officers would have recognized that Comrie had written several times to the agency and that his last letter had not been so complimentary. Only six month before, Comrie had complained that the computation of the Mathematical Tables Project “seems to me extravagant, and to savour of computing gone amok.”
⁸⁰
Comrie's December telegram reached the Mathematical Tables Project only because Comrie sent a copy of it to Arnold Lowan and Lowan distributed copies of the message to all of his supporters.
⁸¹

In fact, when Comrie sent his urgent telegram, two efforts were under way to commandeer the project for military work. The first came from the navy, which had discovered that there were no other large computing laboratories in the United States that were prepared to handle LORAN computations. The woman Rhodes recalls visiting the project was likely Regina Schlachter, a lieutenant (junior grade) attached to the navy's Hydrographic Office. Schlachter visited the Mathematical Tables Project one month before Comrie sent his telegram to Roosevelt. Schlachter examined the operations of the project, determined that the calculations could be secured, and recommended that the navy claim the services of one mathematician, Milton Abramowitz, sixty calculating machines, ten typists, and forty-nine computers.
⁸²

With the resources of the Mathematical Tables Project, Lt. Schlachter organized a new computing office, named the New York Hydrographic Project. She found space for the group at the Hudson Terminal Building, a complex of offices in Lower Manhattan that was later redeveloped into the World Trade Center. Workers from the nearby Brooklyn Navy Yard cleared the rooms, reinforced the doors, and installed a safe. At the request of the navy, Lt. Schlachter installed a twenty-four-hour guard at the facility. She deferred all mathematical questions to Abramowitz, who was assisted by an MIT professor with a reserve commission.
⁸³

The hydrographic computers were a small cross section of New York's population. There were twenty-six men and twenty-three women. The group included a Weinberg, a Sinclair, a Nabokov, an O'Brien, a Dalrimple, and a Cordova. Some of the computers came from Brooklyn, some
from the Bronx, and a few from the neighborhood of Harlem. The navy made no attempt to investigate the loyalty of these workers. Instead, it tried to isolate the computers and prevent them from having access to the final tables. The computing sheets made no reference to LORAN or navigation. There was no indication of time, of radio frequencies, or even of longitude and latitude. When computers resigned, the navy tended to replace them with the wives of servicemen, reasoning that women with a personal stake in the success of the military would be unlikely to betray the office.
⁸⁴

The actions of the navy were matched by a decision from the National Defense Research Committee. In the fall of 1942, the committee was in the middle of a major reorganization, which expanded the number of divisions. During this effort, the committee created a new division called the Applied Mathematics Panel and put this group under the leadership of Rockefeller Foundation mathematician Warren Weaver.
⁸⁵
This panel was the offspring of the division devoted to fire control, a division that was also led by Weaver. Weaver had argued for the Applied Mathematics Panel because “the demands to carry out analytical studies kept increasing rapidly”
⁸⁶
and because, without it, new devices could not be designed, tested, manufactured, and deployed “in time to affect the conduct of the war.”
⁸⁷
In general, the “analytical studies” were expansions of ballistics work, “the mathematical analysis of certain fundamental problems.”
⁸⁸
These studies developed mathematical models for bombing runs, antiaircraft fire, shock wave propagation, and other aspects of weapons operations.

Weaver recruited Thornton Fry of Bell Telephone Laboratories to be the vice-chair of the committee and asked Oswald Veblen to bring his experience from the First World War. In all, about ten mathematicians served on the Applied Mathematics Panel, including Princeton professors S. S. Wilks (1906–1964) and Marston Morse (1892–1977) and New York University mathematician Richard Courant (1888–1972). For a time, the brother of Oswald Veblen's First World War colleague Forest Ray Moulton served as staff to the committee.
⁸⁹
At the first meeting of the panel in January 1943, Weaver opened the discussions by identifying computing as “a large and important need” and argued that the work of the panel would “doubtless involve several broad contracts with groups such as the Lowan WPA Computing Group, the Thomas J. Watson Astronomical Computing Bureau, the Computing Center at MIT, etc.”
⁹⁰

The Thomas J. Watson Astronomical Computing Bureau was the new name for Wallace Eckert's old laboratory at Columbia. The Computing Center at MIT was a small group of computers working directly with Phil Morse.
⁹¹
As both of these organizations were university research labs, they were obvious candidates for Applied Mathematics Panel contracts.
The Mathematical Tables Project was more problematic. Weaver told the first meeting of the panel that he had met with Arnold Lowan and Gertrude Blanch in mid-November, three full weeks before the WPA announced its liquidation. He reported that it was a good discussion and that he came away from the meeting with a better understanding of how the group operated and the kind of work it was able to do. He wanted the panel to take control of the project, though he admitted that no “significant fraction of the group could be cleared” and that the computers would have to be limited to “work of such a general character that could be unclassified.”
⁹²