Friday, February 13, 2009

How the Early Census was Counted

The first commercial data processing machines were punched card tabulating systems. Herman Hollerith (1860-1929) worked at the US Census Bureau during 1879-82. While there he began designing machines that could reduce the labor and time that would be required to process the data that would be collected in the 1890 Census. In 1884, Hollerith applied for his first patent. He proposed to store information in the form of holes punched through a strip of paper. "Holes punched in a strip of paper were sensed by pins or pointers making contact through the holes to a drum. The completion of an electric circuit through a hole advanced a counter on a dial." (G. D. Austrian, Herman Hollerith: Forgotten Giant of Information Processing, 1982, p. 23)

Hollerith switched to punched cards in 1886 and obtained a second patent in 1887. Punched paper cards had previously been used to program silk looms and difference engines. (James Essinger, Jacquard's, 2004) The photograph below shows one of several models of c. 1800 punched card silk looms in the Musée des Tissus in Lyon, France. Also, punched paper rolls had been used in player pianos. The Hollerith Electric Tabulating System consisted of punching, reading, sorting, and tabulating machines.
Early Hollerith Tabulating Machines and Card Readers

The tabulator was a counting machine. It kept a running count of the number of cards with a hole punched in a particular position. It had 40 counters and hence could simultaneously count the number of cards with holes punched in up to 40 positions. An experienced operator could tabulate 50 to 80 cards a minute. There was no printer. The results of a tabulation had read on the counter dials and written down by hand.

Check this site out to see a lot more early office machines and how the census was taken. The 1940 census due to be released in 2012 will be the first one I will be on as a child! Howard.

No comments: