A Brief History of Character Codes

The 1890 U.S. Census and the Birth of a New Character Code

The end of the 19th century saw the creation of another character code, this one invented in the United States for the purpose of tabulating census data. It was created by a young American inventor named Herman Hollerith (1860-1929), who was hired by the U.S. Census Bureau in 1880 as a statistician after graduating from Columbia School of Mines in New York in 1879; and it was to have far reaching effects that lasted into the golden era of mainframe computers in the 1970s. Hollerith was none other than the creator of the Hollerith code, a character code for encoding alphanumeric data on the "punched [or punch] card," which introduced one of the first geek expressions to the American masses--"do not fold, spindle, or mutilate," an expression that left many Americans with the impression that computers were soon to take control of their society.

Interestingly, by the time Hollerith applied it to the recording of census data in the 1880s, the punched card was very "old technology." Punched cards were originally devised around 1800 by Joseph-Marie Jacquard (1752-1834), a French inventor who employed them in a new loom for weaving patterns into fabrics. His invention, patented in 1804, later enabled the creation of the fully automatic loom, and hence the textile industry as we know it today. Moreover, Hollerith was not the first to consider Jacquard's invention for use with computers--Charles Babbage (1792-1871) in England planned to use them with his "analytical engine," a giant mechanical computer that was never successfully implemented. On top of that, Hollerith didn't consider the punched card for data input until he had experimented and failed with data encoded on strips of paper, which unfortunately had a tendency to rip during processing.

To get around the problem of data paper strips ripping, Hollerith devised what was to become the computer punched card (click here to view an image of a Hollerith punched card). In the form in which it was later standardized, it was a 186mm x 82mm stiffened card with a clipped upper left-hand corner that had with 12 rows (numbered 12, 11, 10, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, top to bottom) and 80 columns (numbered 1 through 80, left to right). Each column represented a single alphanumeric character or symbol. As the cards were fed through a reader/calculator, called a "tabulating machine," pins passed through the positions where holes were punched completing an electrical circuit and subsequently registered a value. Although this sounds like a very primitive and limited system, it was very powerful and very efficient. The 1880 census in the U.S., which was done by hand by humans, took seven years to complete, making the data obtained from it almost useless. When Hollerith's "tabulating machines" were employed for the 1890 census, it took the Census Bureau just six weeks to complete the computations with more detailed data than before, and it resulted in a savings of $5 million, since the Census Bureau's personnel costs were greatly reduced.