the beginning

The Punched Card

Hollerith’s punched card was a leap in information technology which can be considered the birth of the data processing industry

In 1890, the U.S. Government had a big problem on its hands. It had taken seven years to tabulate all the data for the 1880 census which had only asked 25 questions of each of 50,189,209 people2. Ten years later, six more states had joined the union and a massive influx of immigrants from Europe had increased the population by another twelve million. It was estimated that it would take thirteen years to complete the 1890 census. They would have collected the data for the 1900 census three years before knowing the results of the 1890 census rendering it useless.

Herman Hollerith1, a graduate engineer from MIT and an employee of the U.S. Census Bureau, was responsible for inventing the machine that would save the census. He proposed that the census data be recorded on punched cards and a machine he called the “Tabulator” would read the cards and count the results. Hollerith’s new invention completed the work of the 1890 census on December 12, 1980, tabulating 62,979,7662 people, in only three months.

The census department was processing 590,000 people a month in the 1880 census. Hollerith gave them the ability to process 20 million a month; a 3500% improvement.

The world was never the same again.

The Jacquard Loom

Hollerith did not invent the punched card. In 1801, Joseph Marie Jacquard invented a mechanism to control the Bolus hooks of looms using a series of cards with holes to control levers which in turn controlled the pattern of woven cloth. The cards were strung together into a loop so that they could be read one-by-one by the machine. Although Jacquard’s invention looks clumsy by today’s standards, it actually worked quite well.

Jacquard’s system did not use holes to represent “information” in the formal sense. The idea of using the hole to mean something was borrowed from railway tickets in which conductors would punch holes through the passengers’ destinations which were pre-printed on the tickets. In an inspired moment, Hollerith combined Jacquard’s concept of machine-readable punched cards and the railways’ ticket punching scheme together. The result would be a card with holes punched in special locations to convey information. Hollerith thought that this could help solve the census problem. All the information for an individual would be contained on a one card, with holes punched into specific places to represent different pieces of information about that person.

Card vs. Paper TapeFor purposes of the census, Hollerith thought of using paper tape instead of cards since it would be easier to feed a paper tape into a machine than a series of cards. Paper tape technology had been well developed by 1890 and off-the-shelf components would have been readily available. The problem with paper tape was that mistakes could not be corrected. Should a hole be punched in the wrong place, the operator couldn’t easily correct the mistake since cutting out the bad section and splicing in a replacement section would be too time consuming to be practical. Using cards was a better solution as the bad card could simply be discarded and replaced with another.

It’s not clear if Hollerith realized that he would one day invent machines to sort his cards – something that would have proved impossible had he gone with paper tape.

The First Punched Card

Hollerith designed a card with a 12 x 20 grid positions3 onto which round holes were punched. To save costs, the cards were the same dimensions as the 1887 U.S. paper currency. The bottom left corner was cut diagonally so that upside-down and/or backward cards could be found. The grid was arranged more like a business form; not in the columnar format that would follow in later years. This made transcribing census information onto the cards easier for the human operators who laboriously punched the cards one hole at a time. A good operator could prepare 500 cards a day.

The Machines

Hollerith then thought that a machine could be built that could process the cards, with pins to test if there each hole position, and use the electrical current when a hole was found to activate a counter. Basically, the machine was nothing more than a device to count holes.

In 1890, there was no mechanical card feed for Hollerith’s machine. The cards were manually put into the machine. The operator placed the cards, one at a time, into a device that looked like a wine press. 240 pins were lowered when the operator pushed down on the lever. Where there were holes, the pins made electrical contact, and the current drove counting dials. When the operator let go of the lever, a spring raised the press and the operator removed the card.

Hollerith’s Original Tabulator

Hollerith invented a Sorting Box as an attachment to the Tabulating machine. In addition to activating the counters, the compartment lids of the Sorting Box were opened electrically. The operator simple deposited the card just read into the open compartment and closed the lid.

The Birth of IBM

Hollerith commercialized his invention in 1896, forming a company that would evolve into a company called Computing Tabulating Recording Corporation in 1911. A later CEO, Thomas J. Watson, would change this name to International Business Machines in 1924.

By the turn of the century, there was a significant advance was in how the information was to be arranged on the cards. Instead of grouping holes in a two dimensional layout similar to a business form, the cards were now treated as one dimensional (one number per column) and the cards could now be processed by simple, general purpose machines.

By 1901, Hollerith’s company began shipping an automatic card sorter. Sensing one of the holes in a column selected by the operator, the cards dropped into one of twelve pockets4.

In 1906, Hollerith patented a device to automate punching holes on the card. There were now 45 columns each with 12 rows of holes. Twelve keys were pressed to punch a hole in one of the 12 positions of a column, and the device automatically advanced the card to the next column. (Alphabetic codes were added in 1933 when IBM unveiled the IBM 032 card punch.)

Although this new scheme actually lowered the amount of information that could be encoded onto a single card, it was much more general purpose. Although the size of the cards remained the same, the density of holes increased. By 1928, precise rectangular holes were now possible and the standard 80 column card was born. It would remain the de-facto standard until the mid 1970s when IBM launched the short-lived 96-column card.

An typical 80-column computer card

Hollerith’s 1890 Tabulator underwent changes as well. The original machine was hard wired to tabulate the data for the 1890 census. It couldn’t do anything else. By 1906, Hollerith’s company engineered the first automated feed tabulator what could read 150 cards a minute complete with a wiring panel that allowed it to be re-wired to process other information.

The tabulators evolved into general purpose accounting machines. Programmed with wiring panels, they could read cards, perform rudimentary logic, and output the results on typewriter devices or onto punched cards.

The Evolution of the Machines

By the 1930s, a whole new industry had evolved. Even without having computers at their disposal, it was now possible to collect, store, and process all sorts of data using equipment that could do simple things like sort cards, read and punch new cards, and total columns of batches of cards.

Cards were used just about anywhere were large volumes of information had to be processed. The concept of employing general purpose machines that could tabulate, sort, collate, accumulate, perform simple logic, and print the results, was a turning point in how information was processed. A whole new industry was born. The early data processors were able to invent all sorts of new business processes simply based on cards being run through these remarkably simple machines, long before computers were invented.

next …

1U.S. Census Bureau
21930 Census Resources for Genealogists
3Columbia University Information Technology
4International Business Machines