Read When Computers Were Human Online
Authors: David Alan Grier
During that first summer of the fair, Gertrude Blanch was still attempting to accumulate the information that she needed for her computing plans. Like others before her, she was learning that there was no single literature of computation. When she needed some mathematical theorem or technical analysis, she would send a junior member of the planning committee to the New York Engineering Societies Library with instructions to scan through some collection of journals, page by page if necessary. The fact that many on the planning committee were immigrants or the children of immigrants simplified such searches, as much useful material could be found in foreign language publications, such as the
Archiv der Mathematik und Physik
, the
Mémoires couronnés et autres mémoires publiés par l'Académie Royale des Sciences, des Lettres et des Beaux-Arts de Belgique
, and the
Wissenschaftliche Schriften des Donetz-Tecknikums des Genossen Artjem zu Stalin
.
The one organization that might have been able to assist the Mathematical Tables Project with its library searches, the Subcommittee on the Bibliography of Mathematical Tables and Other Aids to Computation of the National Research Council, was still moribund and unfocused. By the spring of 1939, the leaders of the National Research Council had lost all hope that A. A. Bennett would be an effective leader of the group. After the initial flurry of activity in 1935 and 1936, Bennett had all but abandoned MTAC. His communications with the National Research Council had became a litany of excuses. “Unexpected and extended interruptions have retarded the work in a way that was not anticipated,” he reported that April.
7
Most of these “unexpected interruptions” had come from the Aberdeen Proving Ground, where Bennett served as a consultant. The proving ground had expanded its computing facility with a differential analyzer, the machine that had been invented to solve differential equations. “During the last part of the year,” wrote the base commander, “the analyzer has been used in the computation of two firing tables with gratifying results.”
8
As gratifying as such results may have been, they did not suggest that the new computing machine would replace the human computers or eliminate all work for A. A. Bennett. The analyzer “saves a great amount of labor when a group of related trajectories are to be computed,” reported one of Bennett's colleagues, but the device was sensitive
and suffered from “mechanical inaccuracies.” The adjustment of the analyzer was “a delicate matter, requiring so much time that for a single trajectory [it was] more economical to compute in the usual way.”
9
Sometime that spring, the leaders of the National Research Council quietly asked Bennett to resign his chairmanship. In his stead, they appointed Raymond Claire Archibald, who was, like Bennett, a professor of mathematics at Brown University. Archibald was a tall, imposing figure, filled with energy and topped by a head of hair that had retained its red color. He was a Canadian by birth and a distant cousin of Simon Newcomb, a connection that gave him great pride.
10
He was also unmarried, a fact that he prominently displayed in his biographies and resumes.
11
The council had twice passed over Archibald when it had sought a chair for MTAC, but in 1939, it was willing to accept anyone who might actually complete a bibliography. Archibald had already proven that he could be a leader of mathematicians, though not quite a leader with the stature of Veblen or even A. A. Bennett. During the First World
War, Archibald had edited the
American Mathematical Monthly
, a prime job in ordinary times but one that seemed small compared to the experience of the Aberdeen veterans. Following the war, he had held several minor positions in the American Mathematical Society and, in the process, acquired the reputation of being difficult. He had once drawn a quarrel from Oswald Veblen over plans for financing the American Mathematical Society. “I gather from your letter,” Veblen had scolded, “that you have not understood my position in the matter. But on re-reading my [note to you], I don't see how I can make it clearer. So I fear all I can do is to ask you to reread the last paragraph of that letter.”
12
36. R. C. Archibald in his office at Brown University
When R. C. Archibald accepted the chair of MTAC, he put all of his considerable personality into the job. Within days of his appointment, he had written to the committee, asking members for “reactions, advice, comments and suggestions on various matters.” He explained that the committee would act as a “clearing-house of information about Tables,” that it would cooperate with England's Mathematical Tables Committee in order “to avoid duplication of effort,” and that it might “take steps toward initiating the development of other tables which were thought to be desirable.”
13
He proposed to establish a broad and inclusive committee that could direct any kind of computational work. Before the summer had ended, he had reorganized the group along lines that had been developed by the Mathematical Tables Committee of the British Association for the Advancement of Science, a structure that split the broad literature of mathematical tables into twenty-one different classes.
T
YPOLOGY OF
M
ATHEMATICAL
T
ABLES
B
RITISH
A
SSOCIATION FOR THE
A
DVANCEMENT OF
S
CIENCE
A. | Arithmetical Tables |
B. | Tables of Powers |
C. | Logarithms |
D. | Circular Functions |
E. | Hyperbolic and Exponential Functions |
F. | Theory of Numbers |
G. | Higher Algebra |
H. | Tables for Numerical Solution of Equations |
I. | Tables Connected with Finite Differences |
J. | Summation of Series |
K. | Statistical Tables |
L. | Higher Mathematical Functions |
M. | Integral Tables |
N. | Interest and Investment |
O. | Actuarial Tables |
P. | Engineering Tables |
Q. | Astronomical Tables |
R. | Geodetic Tables |
S. | Physical Tables |
T. | Critical Tables of Chemistry |
U. | Navigation Tables |
Many of these categories had little in common. Integral tables (M) involved no calculation at all. Interest tables and actuarial tables (N and O) were prepared for businesses, not laboratories. Many engineering tables (P) contained collections of data that had been gathered from experiments. In addition to these categories, Archibald proposed one final division, Z, that would not deal with tables at all but would prepare the bibliography of calculating machines. For each division on the list, he intended to create a small section of four or five members. The chairs of the sections plus Archibald would form an executive group.
The plan would create an unusually large committee for the National Research Council, and it did not take long for the organization's staff to object to Archibald's plans. “There is nothing in the by-laws or procedures of the Council to contravene what Professor Archibald proposes,” lamented the council secretary. “However, I cannot help feeling that making a difference between classes of membership in this way introduces a rather invidious distinction.”
14
Grateful for any action within the committee, the council was willing to let Archibald proceed, even though one scientist on the council compared the new MTAC structure to the myriad agencies of Roosevelt's New Deal and quipped that Archibald “apparently has caught the alphabet fever from the Government.”
15
Undeterred by the objections of the council, Archibald began making the appointments to the committee. He assigned H. T. Davis the chair of Section E, which dealt with material covered in Davis's encyclopedia of functions. L. J. Comrie was given the leadership of the computing machine group, Section Z. To that section, Archibald added George Stibitz of Bell Telephone Laboratories. For Section K, which dealt with statistical tables, he offered the chair to W. W. Edwards Deming (1900â1993), a U.S. Department of Agriculture statistician and a former student at Karl Pearson's Biometrics Laboratory. To catalog the literature of actuarial tables, Archibald turned to the New York insurance industry.
16
He also attempted to expand the international presence on the committee by recruiting Tadahiko Kubota of Japan's Tohoku University.
17
For the section on astronomical tables, Section Q, Archibald turned to the obvious choice, Wallace Eckert of the Columbia University Astronomical Computing Bureau. “Many thanks for considering me in connection with this Committee,” wrote Eckert in reply to Archibald. “I believe the work important, and am willing to do all I can to make the [committee] a success.”
18
Other than L. J. Comrie, Eckert was the only section leader drawn from the astronomical community, a discipline that
had once defined scientific computation. Eckert's little section also contained the only female member of MTAC, a Naval Observatory computer named Charlotte Krampe (1904â1969).
19
By the fall of 1939, Archibald had appointed the core members of his committee and had raised a $15,000 publication fund from the Rockefeller Foundation.
20
At first, this money seemed to be a sign of Archibald's success, but it brought him into conflict with the leadership of the National Research Council. The council leadership believed that they had ultimate control over the funds, as MTAC was a subcommittee of their organization. Archibald claimed that he alone could decide how the money was spent. The disagreement was entirely academic until Archibald published his first bibliography and sent copies of it to the council. In the box, he included an invoice for $61.73.
21
“I cannot understand the item at the bottom of your second page,” wrote one leader of the council, Luther Eisenhart (1875â1965), “unless you mean that since the N.R.C. will use these copies for distribution the N.R.C. should pay the fund for them. This would raise an issue which I have never heard of before.”
22
Whether or not the council had heard of the issue was of little matter to Archibald. He argued that the Rockefeller Foundation funds were his responsibility and that he was required to recover every dime he spent on publication, including the value of the books sent to the National Research Council.
23
Fundamentally, Eisenhart was sympathetic to Archibald. He, too, was an applied mathematician and understood that $15,000 publication funds were rarely given for mathematical research.
24
Yet Archibald's claims verged on insubordination, if not outright mutiny. He considered ousting his committee chair, but he confided to a colleague that “I do not know whom I could possibly secure to replace him and who would be willing and able to do the work in the way in which he has been doing it.”
25
The conflict waxed and waned for nearly two months as Archibald pushed his claim on the council and Eisenhart deflected his demands. Archibald seemed to grasp, though perhaps for misguided reasons, that he had a secure hold on the MTAC chair. For his part, Eisenhart wanted to resolve the conflict without showing any emotion and without weakening the discipline of the other subcommittees. The resolution came when Archibald, frustrated and impatient with the exchange, accused Eisenhart of using the “insidious political methods of the Roosevelt administration.”
26
Eisenhart may actually have chuckled at receiving this overblown piece of rhetoric, but to Archibald, they were clearly fighting words. The MTAC chair held his ground until he was moved by the still, small voice of conscience. “I beg,” he wrote to Eisenhart three days later, “that you regard the last five lines of the fourth paragraph of my letter to you of July 29 as never written, that we may be âas before.'”
27
Eisenhart
accepted the apology, confirmed the council's authority over MTAC, but spared Archibald's ego by paying the $61.73.
28
It is easy to view the fight between Archibald and the National Research Council as a clash among strong personalities or the attempt of a maverick scientist to establish a small empire of his own. In the right hands, it might even have been turned into an entertaining evening of musical satire. However, the event captures the state of computing in 1939 and suggests that the discipline had reached a critical point in its history. With only a few exceptions, computing laboratories had always been under the control of scientists whose interests lay elsewhere. The directors of computing groups had been astronomers, surveyors, electrical engineers, ballistics engineers, physicists, statisticians, meteorologists, and economists. Only people like H. T. Davis and L. J. Comrie could claim to have computing as their primary interest, and of these two, Comrie had been trained as an astronomer.