Punched cards were a very popular medium for input/output and data storage in the early period of computer usage. Data was stored in them by the presence, or absence, of holes punched in pre-determined locations.
They were actually a hangover from a pre-electronic stage of data processing; Herman Hollerith pioneered the use of cards for data storage for the 1890 U.S. census, and IBM became a world-wide colossus before World War II on its dominance of card processing.
Physically, they were stiff, thin cardboard, typically slightly less than .01 inches thick. A number of different formats were used over time, but the most popular by far was IBM's format, introduced in 1928: they were 7-3⁄8 by 3-1⁄4 inches in size, with rectangular holes in 80 columns of 12 rows.
A card could hold up to 80 alphanumeric characters, one per column: numbers were indicated by a single punch in rows 1-10; letters were indicated by a punch in row 10/11/12, plus a punch in rows 1-9. Cards could also hold binary data, with more than two holes per column; by convention, a 7 and 9 punch in the first column indicated that the rest of the card held binary data.
In the pre-electronic era, wire brushes were used to sense the presence or absence of a hole; later on, optical sensors were common.