Howard Aiken and the Harvard Mark I

Abstract

Howard Hathaway Aiken (1900–1973) was a Harvard physicist who in 1937 proposed a large program-controlled calculator. IBM built it in Endicott and delivered it to Harvard in 1944, where it ran as the Harvard Mark I (IBM’s name: Automatic Sequence Controlled Calculator, ASCC). It was the first American digital computer, if relay machines count, and the quarrel over who got the credit led IBM to build its own giant. Aiken went on to build the Mark II, III and IV, hosted the first large conference on digital computers in 1947, and started one of the first university programmes in computing.

A Physicist with Equations

Aiken was born in Hoboken, New Jersey, on 8 March 1900 and grew up in Indianapolis. He took a degree in electrical engineering at the University of Wisconsin in 1923 and worked as an engineer before returning to study at Harvard in the 1930s, where he took a master’s in physics in 1937 and a doctorate in 1939. His thesis required solving differential equations by hand, and by 1937 he was thinking about a machine that could do such work under program control.

The calculating-machine maker Monroe was not interested. Aiken wrote a longer proposal, which cited Charles Babbage and Herman Hollerith, and sent it to IBM, the leader in punched-card equipment. With Harvard’s support, where he now taught mathematics, the university and IBM agreed on 31 March 1939 to build a large calculator.

The ASCC

IBM engineers built the machine at the company’s plant in Endicott, New York, and moved it to Harvard in February 1944. It was about sixteen metres long, weighed 4.3 tonnes and contained 3,500 relays and 800 kilometres of wire, in a chrome-and-glass case styled by the industrial designer Norman Bel Geddes. Its store held 72 decimal numbers of 23 digits, plus 60 settable constants. Instructions came from punched paper tape; results went out on punched cards and electric typewriters. It managed three additions a second, and a multiplication took six seconds.

The machine went to work before its official debut. On 29 March 1944 John von Neumann used it for calculations for the secret atomic bomb project, and from August it worked mostly on weapons problems. It stood at Harvard but formally belonged to the US Navy; Aiken and his assistant Grace Hopper were both naval reserve officers. Konrad Zuse’s Z3 had been running in Berlin since 1941, but nobody in America knew about it; Zuse, by contrast, heard of the Harvard machine during the war, when the daughter of his bookkeeper, a secretary in German intelligence, saw a photograph with the words “Sequence Controlled Calculator” on it (see Konrad Zuse and the Z3).

Who Built It

Harvard received the machine on 7 August 1944. The New York Times covered it that day and ran a second piece the next day, “Aiken’s Calculator,” that credited Commander Howard Aiken and IBM engineers. A newsreel showed sailors in white uniform clambering around the machine and Hopper operating it. Harvard called it the Mark I, and the press followed Harvard: a Popular Science feature of October 1944 put Harvard and Aiken in front and mentioned IBM only briefly. Thomas J. Watson, whose company had put about $200,000 into the machine, disliked such coverage. His answer was the SSEC, a giant of IBM’s own that went into service at the New York headquarters in 1948 (see IBM: The Company).

The Mark I ran around the clock into the 1950s and was partly dismantled in 1959. Some sections remain at Harvard; a control unit is on show at the Heinz Nixdorf MuseumsForum in Paderborn.

The 1947 Symposium

From 7 to 10 January 1947 Harvard and the Navy’s Bureau of Ordnance held a Symposium on Large-Scale Digital Calculating Machinery in Aiken’s new Computation Laboratory. It drew 336 participants, four times as many as a Boston conference on computing in October 1945. After the welcome addresses the first session turned to the Mark II, which Aiken’s team was finishing for the Navy, and the first technical talk was given by Richard Henry Babbage, a Canadian journalist and great-grandson of Charles Babbage, on his great-grandfather’s engines. Later speakers covered the Mark I and II, ENIAC and the Bell Labs relay machines of George Stibitz, who also spoke, as well as storage (Jay Forrester on electrostatic memory, Jan Rajchman on the Selectron), numerical methods (Richard Courant, Hans Rademacher, Wassily Leontief), and programming (Herman Goldstine) and the EDVAC (John Mauchly). The audience included Claude Shannon, George Dantzig, Richard von Mises, Vladimir Zworykin, Hopper, and the only Englishman present, Alan Turing, who was working on the ACE at the National Physical Laboratory (see Alan Turing and the ACE). The proceedings appeared in 1948, and the Charles Babbage Institute reprinted them in 2003.

Mark II to Mark IV

Aiken built three more machines, all without IBM. The Mark II, another relay machine, was completed for the Navy in 1947; the Mark III of 1950 used some vacuum tubes, and the Mark IV was fully electronic, with a magnetic drum and magnetic-core memory. An Wang worked on the Mark IV after joining the Computation Laboratory (see Dead End: Wang Labs).

Teaching Computing

In 1947 Aiken introduced a master’s programme in what would become computer science, nearly a decade before similar programmes appeared at other universities, and the Computation Laboratory trained doctoral students who became prominent in the field, among them Gerrit Blaauw, Fred Brooks, Kenneth Iverson and Gerard Salton (see The Birth of Computer Science as a Discipline). In June 1959 he presided over UNESCO’s International Conference on Information Processing in Paris, which led to the founding of IFIP.

Aiken retired from Harvard at 60, moved to Fort Lauderdale, set up a consulting firm and taught at the University of Miami. He died on 14 March 1973 in St. Louis, while on a business trip, a few weeks after recording a long oral-history interview for the Smithsonian.