Herman Hollerith
In the 19th century, the United States was drowning in data. The Census Bureau struggled to keep pace with an exploding population. Herman Hollerith would create a machine that addressed the challenge of processing these mountains of forms.
In 1879, fresh out of Columbia School of Mines, he joined the Census Bureau. He met Dr. John Shaw Billings, who headed the vital statistics division, and who suggested the idea of mechanizing data processing. Indeed, the stakes became obvious when looking at the 1880 census. Fifty million Americans to count. Operators laboriously processing twenty characteristics per minute with rudimentary tools. The figures were not published until 1889, barely a year before the next census. The situation was untenable.
Hollerith then left the Bureau to follow General Francis Walker to MIT. Having become an instructor in mechanical engineering, he began developing his first prototypes. He drew inspiration from Colonel Charles W. Seaton's work, who had created a machine using a roll-feed system comparable to that of a mechanical piano. But Hollerith would take a different direction.
After this academic interlude, he returned to Washington, joined the Patent Office, then established himself as an independent expert. This experience gave him the keys to protect his future inventions. He first worked on electric brakes for railroads, outperforming George Westinghouse during tests in 1887. But the examiners doubted electricity's reliability as a power source. No matter.
His data processing system relied on punched cards measuring 6.5 × 3.25 inches. The operator, reading census forms, could encode up to 17 different characteristics among 240 possible positions. Standard codes, such as those for districts, were preset and reproduced on entire stacks of cards.
The tabulating machine revealed Hollerith's ingenuity. In the upper jaw, rows of spring-loaded pins; in the lower jaw, tiny cups of mercury. The pins, connected to an electrical current, faced cups linked to 40 counters. The employee inserted a card into the press, closed the jaws, and the pins passing through the holes plunged into the mercury. The current flowed to the electromagnetic counters, advancing each relevant needle. A complete rotation of the units needle activated the hundreds needle. The card was then placed in a compartment of the sorter before moving to the next one.
The numbers spoke for themselves: 700 punched cards per day per operator, 7,000 cards processed daily by a machine user, 250 characteristics analyzed per minute. Through relays, the system performed more sophisticated analyses, such as identifying "white men born abroad".
The major trial by fire came with the 1890 census. The Bureau ordered fifty machines to count nearly 63 million Americans. Such success that the administration would use improved versions of the system for six decades, until the arrival of UNIVAC I.
Hollerith's influence spread beyond the census. Statistical services, insurance companies, railroads—all wanted this technology that drastically reduced data processing times. In a rapidly changing industrial society, it addressed a critical need. Moreover, some technical choices he made had surprising longevity. His punched card format matched the dimensions of banknotes of the era, simply because filing cabinets adapted to them already existed. His positional coding system, designed for the 1901 agricultural census, would influence IBM and other companies through the 1970s.
The patents filed by Hollerith were the forerunners of the data processing industry. For sixty years, his machine bridged the gap between manual processing and electronic computing. It transformed how we manage information while initiating the techniques of the modern computer industry. An invention born from a very concrete problem.