Mark sensing
Pencil marks read electrically, on the grounds that graphite conducts and keypunch operators draw a salary.
The idea came from Reynold B. Johnson, a high-school teacher tired of marking exams, whom IBM hired in 1934 after buying his invention. A card carried printed boxes, and whoever filled it in used an electrographic pencil with a high graphite content. Inside the machine, metal brushes swept across the positions, and wherever there was graphite the circuit closed. A companion unit then punched real holes at the matching positions, turning the sheet into an ordinary punched card that the rest of the machine room understood.
The point was to move data capture out of the machine room. A stockroom clerk, a meter reader or an exam candidate could produce machine-readable data directly, with no keypunch operator standing in between — as much a payroll question as an accuracy one in that era. The IBM 805 test scoring machine arrived in 1937, and the family resemblance to today's answer sheets is obvious. The difference is that reading is now optical. The graphite is mostly tradition.