Keypunch
A keypunch is a mechanical or electromechanical device used to encode data by punching rectangular holes into stiff paper cards at predetermined positions corresponding to alphanumeric characters, operated via a typewriter-like keyboard that advances the card column by column after each keystroke.[1] The technology originated in the late 19th century, when inventor Herman Hollerith adapted punched cards—initially inspired by Jacquard loom control mechanisms—for use in the 1890 U.S. Census tabulating machines, enabling efficient mechanical data processing and marking the birth of modern information technology.[2] Hollerith's system used 24-column cards with circular holes, but by 1930, the standardized 80-column rectangular-hole format emerged, measuring 7⅜ by 3¼ inches and capable of storing about 80 bytes of data per card, with each hole position representing a specific character via a 12-row encoding scheme.[2] IBM, originally the Tabulating Machine Company founded by Hollerith in 1896, became the dominant manufacturer, producing millions of cards daily by the 1930s on high-speed rotary presses and driving significant revenue—accounting for 20% of sales and 30% of profits by the mid-1950s.[2] Keypunch machines evolved from early manual models like the IBM 001 (circa 1901) to more advanced versions such as the IBM 026 (introduced in 1949), which featured limited programmability via a control drum for automating repetitive data entry, and the IBM 029 (released in 1964 alongside the System/360 mainframe), a quieter, more reliable model with enhanced character support including symbols like parentheses and a U-shaped card path for efficient handling of 80-column cards.[3][4] These devices were essential for data preparation in business, government, and scientific applications, allowing operators—often women in dedicated keypunch rooms—to transcribe source documents onto cards that were then read by tabulators or early computers at speeds up to 1,000 cards per minute via machines like the IBM 2540.[4][2] By the mid-20th century, keypunches facilitated offline programming and data storage for mainframes, as cards provided a portable, inexpensive medium despite limitations like error-prone manual punching (requiring full card repunching for corrections) and vulnerability to damage, famously encapsulated in the cultural admonition "Do not fold, spindle, or mutilate."[2][4] Their widespread use persisted into the 1970s, supporting applications from payroll and inventory to scientific computations, but they were gradually supplanted in the 1950s–1980s by magnetic tape, disks, and direct keyboard input due to higher capacity, speed, and reduced labor costs, with the IBM 029 remaining available until 1984.[2][4]Principles and Technology
Basic Operation and Components
A keypunch machine consists of several core components designed to facilitate the precise encoding of data onto punch cards. The keyboard, resembling a typewriter in layout, features keys arranged for alphanumeric input, including digits, letters, punctuation, and special symbols, with some models supporting dual-purpose keys that shift between numeric and alphabetic functions via a mechanism like a numeric shift key.[1][5] Below the keyboard lies the punch mechanism, which employs needles or chisels to create holes in designated positions on the card, ensuring accuracy through a grid-based system where each column accommodates up to 12 possible hole positions.[6][5] The card hopper, typically holding up to 500 blank cards, automatically feeds cards into the punching station one at a time, while the stacker collects the processed cards in an output tray of similar capacity, advancing them after completion.[3][5] Interlocks, such as keyboard locks and full-stacker sensors, prevent errors by inhibiting multiple simultaneous key presses or halting feed when the output is full.[5] The punching process begins with the operator inserting a card manually or via the hopper, positioning it under alignment guides for proper orientation. Depressing a key activates a mechanical linkage or solenoid that drives the punch tool to perforate the card at the corresponding row and column, with the machine automatically advancing the card to the next column position after each punch.[1][6] This step repeats across the card's columns—typically 80 in standard formats—until the end is reached, at which point the card is ejected to the stacker.[5] Control keys, such as skip or duplicate, allow navigation or repetition without punching, enhancing efficiency while interlocks ensure no overlapping holes.[3] Early keypunch machines relied on manual or hand-operated power sources, where key presses directly drove the mechanical linkages through levers and cams. Later electric models incorporated motors to power the card feed and advancement, along with relays and solenoids for faster, more reliable punch activation, enabling speeds of 9 to 20 columns per second under programmed control.[1][5] Safety and maintenance features include dust covers to protect internal mechanisms from debris, card alignment guides to ensure precise positioning and prevent misalignment, and a chip box to collect punched chads. For troubleshooting jams, a pressure-roll release lever allows manual card extraction, while backspace functions permit correction of up to 20 columns without discarding the card.[6][5]Punch Card Encoding and Standards
Punched cards encoded data using the presence or absence of holes in predefined positions, with the Hollerith code serving as the foundational standard for representation in keypunch systems. Developed by Herman Hollerith for the 1890 U.S. Census, this code utilized a 12-row by 80-column format on cards measuring approximately 7 3/8 inches wide by 3 1/4 inches high, employing rectangular holes measuring about 0.110 inches wide by 0.125 inches tall.[7][2] The rows were numbered from top to bottom as 12, 11, 0 through 9, allowing each column to represent one character through single or multiple punches, with no more than three holes per column to prevent mechanical issues in readers.[8] Numeric values were encoded simply with a single punch in the corresponding row (e.g., digit 5 in row 5), while alphabetic characters combined a "zone" punch in rows 12, 11, or 0 with a digit punch in rows 1-9: for example, A as 12-1, J as 11-1, and S as 0-1. Special characters, such as & (12-0) or $ (8-3), used unique combinations, supporting up to 256 patterns in extended versions aligned with ASCII.[7][8] The 80-column layout dominated due to its balance of density and readability, with column centers spaced 0.09375 inches (3/32 inch) apart across the card's punchable area, enabling reliable electrical or optical sensing in tabulating machines.[9] In contrast, the 96-column format, introduced later by IBM, compressed data using smaller round holes (about 0.050 inches in diameter) and tighter 0.046875-inch (3/64 inch) spacing between column centers, arranged in three horizontal bands of 32 six-row positions each for binary-like encoding (rows labeled B, A, 8, 4, 2, 1 from top to bottom in each band).[10][11] This allowed approximately 120 characters per card but required specialized equipment incompatible with standard 80-column readers. Row numbering in both formats followed the top-to-bottom convention, though the 96-column's subdivided bands prioritized bit-level representation over the zone-digit scheme of Hollerith code. Error detection in punched cards relied on manual or mechanical verification rather than inherent digital parity in early standards, with duplicate punching—where identical data fields were recorded twice on the same card—enabling cross-checks during keypunching or sorting to identify discrepancies.[12] Later advancements incorporated parity checks, such as an odd-parity punch in an additional row or column, particularly in 96-column systems and post-1950s readers, to detect single-hole errors during data readout.[13] These methods improved accuracy for high-volume processing but did not correct errors automatically. Variations in hole shape and format addressed mechanical and regional needs: early IBM cards (pre-1930) used round holes for durability in sorters, transitioning to rectangular holes in the 80-column standard to increase density without enlarging the card.[14] Rectangular holes provided cleaner cuts and better electrical contact in brush-based readers, while round holes persisted in competing systems for reduced jamming. International adaptations included the French 90-column cards, produced by companies like Compagnie des Machines Bull, which featured round holes in a two-band layout (45 columns per band with six rows each) to accommodate local tabulating equipment and character sets.[15][9]| Character Type | Example Encoding | Hole Positions (Row Combinations) |
|---|---|---|
| Numeric | 5 | Single punch in row 5 |
| Alphabetic | A | Punches in rows 12 and 1 |
| Alphabetic | S | Punches in rows 0 and 1 |
| Special | & | Punch in row 12 and 0 |
| Special | $ | Punches in rows 8 and 3 |