@Mark Ollig
In 1937, Harvard University physicist Howard H. Aiken envisioned a powerful electromechanical calculating machine to solve complex scientific mathematical problems.
Inspired by Charles Babbage’s Analytical Engine, Aiken aimed to bring his vision to life with the support of Harvard faculty members Professor Emory Leon Chaffee and Harlow Shapley.
In November 1937, Aiken officially submitted his design proposal to IBM, seeking their expertise and resources to construct the automatic calculating device.
After a thorough review, IBM President Thomas J. Watson Sr. approved the project in early 1939, providing complete engineering support and funding.
Construction started that year at IBM’s Endicott, NY, laboratories, led by chief engineer Clair D. Lake, Frank E. Hamilton’s mechanical creativity, Benjamin M. Durfee’s precision assembly skills, and James W. Bryce’s expertise in relay circuitry.
Aiken, Harvard, and IBM established the official name for the device: IBM Automatic Sequence Controlled Calculator (ASCC).
The IBM ASCC spanned 51 feet in length, stood 8 feet tall, and weighed nearly 10,000 pounds.
Enclosed in a stainless-steel and glass frame, the IBM ASCC contained about 750,000 parts, including numerous gears, switches, 3,304 relays, and counters, all working together in coordinated unison, connected by 500 miles of wiring.
The IBM ASCC’s internal computation logic, relays, and control circuits were powered by a 50-volt DC generator, as described in Harvard’s 1946 operations manual, which includes the electrical schematics.
Driving the gears, rotating shafts, and other moving parts was a four-horsepower AC motor, ensuring the machine’s mechanical assemblies remained synchronized.
The IBM ASCC provided a total of 132 storage locations: 72 standard storage registers (accumulators) and 60 constant-register switches (constant dials).
Every register or dial on this machine could hold a decimal number that was as long as 23 digits, and it could show whether the number was positive or negative.
In total, the machine could store 3,036 digits, which was impressive for a computer built before electronic technology was widely used.
This gigantic calculating machine contained six-foot aisles so operators could move alongside it and access its backside cabling and parts.
Programming the IBM ASCC required collaboration among coders, mathematicians, and engineers to convert mathematical problems into machine instructions.
Operators, typically Navy personnel or technicians, loaded program tapes, managed data, and supervised calculations.
The IBM ASCC program instructions were encoded as specific hole patterns on long paper tape, using up to 24 columns per row.
Unlike modern binary codes, the holes in paper tape represented decimal values, register numbers, or machine functions.
Each position on the tape corresponded to a specific operation or register within the IBM ASCC.
The pattern of holes in each row of the tape defined an instruction, specifying which operation to perform and which registers or constants (fixed values set in the IBM ASCC) to use repeatedly during calculations.
The finished rolls of paper tape served as “programs” for the IBM ASCC, which read each row of instructions in sequence, line by line.
Setting up a new calculation required more than just punching holes in paper tape.
Coders, along with operators, would configure physical patch cords and adjust ten-pole switches controlling various circuits, ensuring the data was routed correctly within the computing machine.
Input values were fed into the IBM ASCC using punched cards, while built-in IBM electric typewriters printed data output on paper.
For its time, calculations run using the IBM ASCC were solved remarkably fast.
Addition and subtraction operations took about 0.3 seconds to complete, while a multiplication problem took between three and six seconds.
Calculating division on the IBM ASCC took about 15.3 seconds, nearly four times longer than multiplication, due to the increased number of mechanical cycles required, which put greater demands on the machine’s moving parts.
The IBM ASCC could automatically solve complex scientific problems, including systems of linear equations, by executing lengthy arithmetic sequences.
It performed single sine or logarithm calculations in about a minute, while desk calculators of that era were unable to handle these functions, relying instead on printed tables, slide rules, and approximations.
The completed IBM Automatic Sequence Controlled Calculator was delivered in early 1944 to Harvard University in Cambridge, MA, and installed in the Cruft Physics Lab, where it became operational in May.
Harvard University President James B. Conant formally accepted the IBM ASCC from IBM President Thomas J. Watson Sr. during a public dedication ceremony Aug. 8, 1944.
Soon after, the IBM ASCC became known as the Harvard Mark I.
During WWII, the US Navy Bureau of Ships used the Harvard Mark I for generating gunnery tables, processing reports, and performing military engineering calculations.
The Harvard Mark I played a key role in the Manhattan Project by solving complex mathematical problems related to shock waves and detonation timing for atomic bomb design.
Its precise calculations were used by the engineering teams developing the atomic bombs dropped on Japan in 1945.
“A Manual of Operation for the Automatic Sequence Controlled Calculator,” written in 1946, was edited by Lt. Grace Hopper, who is credited with extending, revising, and writing several chapters.
The manual states that she “more than any other person is responsible for the completion of the book.” It is available on the Internet Archive at https://bit.ly/47bnIT5.
The Harvard Mark I’s moving relays, counters, gears, switches, drive shafts, and sprocket drums advancing the program paper tape generated a rhythmic mechanical symphony, until it was decommissioned in 1959.
You can see and hear the Harvard Mark I, which helped launch the dawn of modern computing, at:https://bit.ly/40Nz384.
