For an older generation of people the sight of a 7 3/8 inch long rectangular card covered in rows of numbers with lots of squares punched out will bring back strong memories. And for those that do remember they will probably then turn to the younger generation around them, tell them how lucky they are now, and regale them with stories about how before we had the luxury of stored memory, floppy disks and the like, early computers had to be laboriously hand-programmed with such antiquated punch cards.
When computers first appeared on the scene back in the early 1950s they were basically little more than glorified calculators. Unable to hold long-term memory, every calculation and action we needed the machine to do had to be told and then as necessary reminded to the computer one instruction at a time. And even back then the punch card system of data processing and interpretation was not of course a new invention, and its origins stretches all the way back to the famous Jacquard loom of 1804.
Joseph Jacquard’s time-saving invention was a device that fitted to an industrial loom to quickly regulate and recall the pattern of weave the machine was intended to make. A long roll of card was pre-punched with a pattern of circular holes that would tell the loom which shuttles to send and when. This could then be rolled up and reloaded on again whenever it was needed. It was a great idea, but apart from use by bus conductors, music boxes, pianola’s and the English mathematician and father of computing Charles Babbage, its simple principle went largely un-tapped for a further 85 years.
Finally a census clerk named Herman Hollerith expanded upon the punch principle to help speed up data processing for the New York Health Service and then for the 1890 US public census. Where the 1880 census had taken 8 years to collate by hand, Hollerith’s new punch card tabulator did the same job in just two. Instead of using a cumbersome long roll of card Hollerith opted instead for individual cards, 7 3/8 inches long by 3 ½ inches high (or interestingly the standard size of most US bank notes at the time) that could be bundled up all together for storage. Hollerith’s punch card system proved a great success allowing him to go on and found a small data processing company now known as IBM. Since the cards were an efficient way of holding and retelling data, it was not surprising that they would be taken up by the fledgling computer industry.
Other variations of punch technology subsequently began to emerge during the early 20th Century such as the 5 and 8 hole wide punch tape that was used extensively by the telegraph and telecommunication sectors. These long paper tikka tapes did find use in many large-scale computer systems during the 1950s and 60s and several other pioneering new methods of storage by using magnetic tapes and disks were also created, but mostly these things were reserved for the most expensive and complex computers around. Instead the dominance of IMB and the sheer practicality of the Hollerith cards made them the first choice for many regular businesses and institutions looking for data processing and basic programming needs. And why rectangular holes? Well there is a valid argument that square holes create a stronger cleaner connection with the reader than circular ones so tend to wear out more slowly. While each different computer system initially required its own individual layout of card, the prevalence of certain programming languages such as Fortran as well as the general growing need for data sharing meant a very standardized 80 column card would become the most common.
As anyone who ever used these punch cards would tell you, they inevitably always created a bit of a head-ache. Once the user had worked out their individual program for the computer, it was time to transfer it onto the cards. This was usually done by qualified key punch operators and often verified afterwards by a supervisor to ensure there were fewer costly mistakes. If mistakes in the code was found that individual cards could then be replaced or the necessary holes carefully refilled.
While automatic card punch machines with their typewriter-style interface were obviously more user friendly, the manual hand punch machines were harder to master. With far less keys numbers were easy enough to punch out, but once you needed regular letters then two or even three keys would need to be pressed simultaneously to get the correct result. It was a painfully precise operation yet for those who did become punch card operators, a solid and in demand career path was possible.
The sharp sounding click of each key would cut the relevant hole on each line and underneath would build a pleasing pile of clippings (or ‘chad’ as it was commonly known). The user would then be left with a tall stack of cards which could be feed into the reader to run their program. Since lines 73-80 of each card were most commonly and intentionally left bank, it was common practice to sequentially hand number each card in the space or mark the outside of the stack with diagonal pen lines to ensure they could be easily put back in order should the pile get mixed up.
By the mid to late 1970s computer punch cards were finally becoming yesterday’s technology and were being replaced by far more user-friendly compact cassettes tapes and then obviously by various forms of core memory. But while no-one probably misses the headache of the punch card system there is undeniably a nostalgia for it all as well. This was a time when programming was as much a physical act as it was a mental exercise, and the sweat and tears of simply creating and feeding the cards must have given a much closer relationship to the whole computer experience.
Photo attribution: IBM Punch Card by Paul Sullivan; IBM 26 Card Machine and IBM hand punch by Marcin Wichary ; Jacquard Loom by Danny Howard ; Holerith Census Machine by Jennifer ; IBM card punch machine by Karl Baron