Harvard Mark I

IBM Automatic Sequence Controlled Calculator (Harvard Mark I)
Company type	Electromechanical computer
Industry	Computing hardware
Founded	1944
Founder	Howard Aiken, IBM engineers
Defunct	1959
Fate	Disassembled; components distributed to IBM, Smithsonian Institution, and Harvard
Headquarters	Cambridge, Massachusetts, United States
Area served	United States
Key people	Howard Aiken, Thomas Watson Sr., John von Neumann, Richard Milton Bloch, Robert Campbell, Grace Hopper
Products	Electromechanical general-purpose computer
Parent	IBM, Harvard University

The Harvard Mark I, also known as the IBM Automatic Sequence Controlled Calculator (ASCC), represented one of the earliest general-purpose electromechanical computers, introduced during the concluding years of World War II.

Among its initial applications was a program executed on 29 March 1944 by John von Neumann, which assisted the Manhattan Project by assessing the practicality of the Implosion-type nuclear weapon. Additionally, the Mark I generated mathematical tables, thereby realizing the longstanding ambition of British pioneer Charles Babbage for his 1837 analytical engine.

The machine earned the nickname "Bessy, the Bessel engine" from its operators, highlighting its frequent use in calculating Bessel functions.

Following its disassembly in 1959, components of the Mark I were distributed among IBM, the Smithsonian Institution, and the Harvard Collection of Historical Scientific Instruments. Harvard exhibited its share at various venues over several decades, most recently moving it in 2021 to the lobby of the new Science and Engineering Complex located in Allston, Massachusetts.

Origins

The inception of the Harvard Mark I dates back to 1935 when Howard Aiken, then pursuing graduate studies in physics at Harvard University, conceived the idea of a large-scale automatic calculator. In November 1937, he presented this concept to IBM as a solution for tackling complex mathematical challenges. Following a feasibility evaluation, IBM chairman Thomas Watson Sr. authorized funding in February 1939.

After facing initial rejection from both the Monroe Calculator Company and Harvard University, Aiken found inspiration upon examining a demonstration model connected to Charles Babbage’s legacy. This led him to integrate design elements from Babbage’s Analytical Engine, effectively bringing those early ideas closer to practical realization while introducing novel enhancements.

In 1941, Aiken temporarily halted development due to his active naval duty during World War II. Meanwhile, IBM undertook construction of the Automatic Sequence Controlled Calculator (ASCC) at their Endicott facility starting in early 1943, delivering it to Harvard by February 1944. The machine began performing calculations for the US Navy Bureau of Ships in May and was formally inaugurated on August 7, 1944.

Although not the very first operational computer, the Mark I was pioneering in automating intricate computations, marking a milestone in computing evolution.

Design and construction

The ASCC was assembled using an array of switches, relays, rotating shafts akin to those in a Pinwheel calculator, and clutches. It incorporated approximately 765,000 electromechanical parts interconnected by hundreds of miles of wiring. The device occupied a substantial volume—measuring long by high by deep—and weighed roughly . Its calculation units were mechanically synchronized and powered through a drive shaft connected to an electric motor that functioned both as the energy source and system clock generator. According to IBM Archives:

The Automatic Sequence Controlled Calculator (Harvard Mark I) was the first functioning machine capable of performing extended computations automatically. Conceived by Harvard’s Dr. Howard Aiken and constructed by IBM engineers in Endicott, New York, the Mark I featured a steel frame stretching 51 feet long and standing 8 feet tall housing interlocking gears, counters, switches, and control circuits. It utilized approximately 500 miles (800 km) of wiring with three million connections, included 3,500 multipole relays with 35,000 contacts, contained 2,225 counters and 1,464 tenpole switches, alongside tiers of 72 adding machines each capable of handling 23 significant digits. It stood as the largest electromechanical calculator within the industry.

The cabinet enclosing the machine was designed by industrial designer Norman Bel Geddes at IBM’s expense. Aiken reportedly expressed dissatisfaction with this expenditure exceeding $50,000, preferring that funds be directed toward acquiring additional computational hardware.

Operation

The Mark I was equipped with 60 groups of 24 switches for manual data input and had the capacity to hold 72 numbers, each consisting of 23 decimal digits. It could carry out three addition or subtraction operations every second; multiplication required six seconds; division took 15.3 seconds; and calculations involving logarithmic or trigonometric functions exceeded one minute.

Instructions were sequentially read from a 24-channel punched paper tape, while input numbers originated from a different tape format that was incompatible. Instructions were not retained in the working memory, which led to assertions that the Mark I influenced the development of the Harvard architecture. However, this is debated since the term surfaced only in the 1970s and is more accurately applied to subsequent Harvard machines such as the Mark III and IV.

The primary sequencing mechanism moved in a single direction, necessitating physically long programs. Program loops were formed by either linking tape ends or unrolling loops; initially, conditional branching was handled manually but later became automated through subroutine calls introduced in 1946.

Among the first programmers were notable figures like Richard Milton Bloch, Robert Campbell, and Grace Hopper. The technical staff operating the machine were unaware of its overall purpose; some had previously worked at IBM before their assignment to the Navy.

Instruction format

The input tape’s 24 channels were segmented into three groups of eight channels each. Every storage location, switch set, and the registers associated with input and output as well as arithmetic units possessed a distinct binary index on the control tape. The first group indicated the binary address of the operation’s result, the second specified the source datum, and the third contained an operation code that identified the instruction to be executed.

Contribution to the Manhattan Project

In 1928, L.J. Comrie pioneered scientific computation using IBM punched-card equipment, particularly for astronomical tables calculated via finite differences—a method initially conceived by Babbage for his Difference Engine. IBM subsequently modified its tabulators for such scientific applications, notably producing the Columbia Difference Tabulator in 1931.

At Los Alamos, John von Neumann’s team employed adapted IBM punched-card machines to study implosion effects. In March 1944, he suggested performing implosion simulations on the Mark I computer and collaborated with Harvard mathematicians to develop appropriate programming.

Calculations completed by the Los Alamos group outpaced those from Cambridge; however, punched-card machines provided results accurate to six decimal places compared to eighteen decimal places achieved by Mark I. Additionally, Mark I utilized smaller integration steps, resulting in substantially improved precision.

Von Neumann joined the Manhattan Project in 1943 and played a key role in extensive computations critical to atomic bomb development. He demonstrated that the implosion mechanism used in both the Trinity test and Fat Man bomb was faster and more effective than earlier gun-type designs.

Aiken and IBM

Aiken released a statement declaring himself the sole creator of the Mark I, acknowledging only James W. Bryce from IBM while excluding contributions from IBM engineers Clair Lake and Frank Hamilton. This omission angered IBM chairman Thomas J. Watson, who begrudgingly attended the dedication ceremony on August 7, 1944. In reaction to this, Aiken opted to pursue future developments independently, resulting in the ASCC becoming widely recognized as the "Harvard Mark I." Concurrently, IBM advanced its technology by developing the Selective Sequence Electronic Calculator (SSEC) to highlight its innovation efforts.

Successors

Aiken progressed with subsequent models: the Harvard Mark II (1947/48), Mark III/ADEC (1949), and Mark IV (1952). The Mark II enhanced the electromechanical relay design of the Mark I. The Mark III featured predominantly electronic components such as vacuum tubes and crystal diodes, combined with mechanical elements like magnetic drum memory and relays for data transmission. The fully electronic Mark IV eliminated remaining mechanical parts by incorporating magnetic core memory. Both the Mark II and Mark III were delivered to the U.S. Navy facility at Dahlgren, Virginia; the Mark IV was constructed for the U.S. Air Force but remained housed at Harvard.

In 1959, the Mark I was dismantled; various parts were exhibited within Harvard’s Science Center as part of its collection of historic scientific instruments before being moved to the Science and Engineering Complex in Allston in July 2021. Additional components had previously been transferred to IBM and the Smithsonian Institution.

References

Publications

in

External links

Oral history interview with Robert Hawkins at Charles Babbage Institute, University of Minnesota. Hawkins recounts details about the Harvard-IBM Mark I collaboration and Howard Aiken’s role.
Oral history interview with Richard M. Bloch at Charles Babbage Institute, discussing his contributions to the Mark I.
Oral history interview with Robert V. D. Campbell at Charles Babbage Institute, addressing Harvard and IBM’s involvement in developing the Mark I.
Photograph identifying machine components: IBM ASCC-Mark I computer framed photograph
IBM Archive: IBM ASCC Reference Room