Sunday, March 11, 2012

Consider the humble punch card. Punch cards had been used in France to control textile looms since the early 1700s; the method was perfected by Joseph Marie Jacquard in 1801 with his Jacquard loom.
     Flash forward to the year 1890, when a man named Herman Hollerith invented a punch card and tabulating machines for that year's United States Census. His census project was so successful that Mr. Hollerith left the government and started The Tabulating Machine Company in 1896. After a series of mergers and name changes, this company became IBM. You may have heard of it.
     Up to the 1970s, the "IBM card" and related machinery was everywhere. The most common card was the IBM 5081, and that part number became the most common term for it -- even across vendors! The punch card was data processing back then.
     The physical characteristics of the card determined how we stored and processed data for decades afterwards. The card was the size of an 1887 United States dollar bill (3.25 inches by 7.375 inches). The reason for that size was simple: when Hollerith worked on the Census, he could get drawers to store the decks of cards from the Department of the Treasury across the street.

Joe Celko, Thinking in Sets: Auxiliary, Temporal, and Virtual Tables in SQL

1 comment:

lo5an said...

It can be a little hard to track down, but I think you might really enjoy the book Data and Reality by Bill Kent.

It digs in to the philosophy of data modeling. It was written in 1978, and you get a nice anachronistic vibe from reading it now, but the issues that it tackles are deep enough that the material is still worth reading.