Excerpted from A Nation Of Numbers by Paul A. Scipione
Little holes in little cards: Hollerith’s helpful handiwork
By Paul A. Scipione
Herman Hollerith was the inventor who first brought new technology to the task of processing and analyzing marketing research data. In doing so, he won a competition sponsored by the U.S. Census Bureau, along with a “prize” that has been variously reported from $1,000 to $50,000. Regardless of the amount, his contract with the U.S. census made him wealthy. Hollerith’s brainstorms were the keypunch card and an electrical/mechanical machine for sorting and counting cards.
Hollerith was born on February 29th, 1860, in Buffalo, N.Y. He graduated from the Columbia University School of Mines in 1879 and then stayed on as an assistant to one of his professors, William Petit Trowbridge, who was helping to plan the 1880 census. Through Trowbridge, Hollerith met John Shaw Billings, a senior advisor on the 1880 census. Over tea one evening, Billings said there ought to be a machine that could take the drudgery out of census-taking and that the Census Bureau was going to sponsor a contest to identify promising new technologies. Hollerith, who had taken a position as an instructor at MIT, had already proven himself inventive by developing a new electro-mechanical air brake for railroads. He entered and won a three-way competition to automate the 1890 census.
Hollerith’s tabulating machine and keypunch cards worked so well in early tests in Baltimore, New Jersey and New York City that he was immediately awarded a nationwide contract. His machine “read” and sorted heavyweight cards, measuring 3-1/4 inches high by 8-5/8 inches wide (non-conductive) on the basis of one or more square holes that were punched under each of 80 columns or “fields” arrayed horizontally across each card from left to right. The upper-left corner of each card was cut diagonally across so that all cards would be oriented the same way as they ran through Hollerith’s machine. Hollerith manually punched one card at a time on a flat punching board/template with small holes drilled into it. A census worker pressed a square-ended stylus down into the tiny hole below the appropriate location in the card to represent the relevant data. Punched cards were then “read” by tiny metal arms in an electrical-mechanical vertical card sorting device.
Although it took less than four (rather than eight) years to tabulate the 1890 census data, clearly there was a need for even faster Hollerith equipment. By the 1910 census, Hollerith introduced a much faster way to punch data cards, a true keypunch machine that looked like an oversized typewriter and introduced drum cards that allowed census clerks to tab to preset data columns. (The keypunch machine that I used to punch the data for my doctoral dissertation in 1972 at Rutgers was nearly identical to the one Hollerith introduced in 1910.) Then the punched cards were fed into a new, improved card sorting machine. It was a 10-to-15-foot-long conveyor belt, under which were a series of bins. A trapdoor opened to allow appropriate keypunch cards to fall through to the bin below, thereby turning a small toothed counting wheel from 0 to 1. Once the counting wheel reached 9, the next turn turned the first two counting wheels to 10, and so forth. After the complete deck of punched cards had been run through the counting machine, a census worker with a clipboard walked along and wrote down the count of cards that had fallen into each bin. The machine operator then had to set all the counting wheels back to 0 and reprogram the combination of punches associated with each bin. For instance, maybe Bin 1 was for males who had graduated from high school and Bin 2 was for males who hadn’t graduated from high school. Bin 3 was for females who had graduated from high school, while Bin 4 was for females who hadn’t finished high school – and so on. There were anywhere from about 15 to 30 such bins below the conveyor belt, which limited the number of tabulations that could be handled during any one run. Although these technologies may seem crude to readers today, they represented a huge improvement in carrying out not only decennial censuses but also the surveys of commercial marketing research analysts.
Based on such Boolean Algebra concepts of if, and, or and not, Hollerith’s strictly analog 1890 counting machine could be considered the first operational forerunner of today’s digital computers, which still use Boolean principles.
On the success of his work on the U.S. census, Hollerith won similar contracts to tabulate national census statistics in Canada, Norway and Austria. He gained rapid fame, winning the prestigious Elliot Crossen Medal from the Franklin Institute in Philadelphia. He was also awarded several valuable patents for his tabulation technologies and these became the basis for his new company, the Tabulating Machine Company (TMC) that he formed in 1896.
In 1911, Hollerith agreed to merge his company with two others, the Computing Scale Company of America and the International Time Recording Company, both of New York, into a new Computer-Tabulating-Recording Company, whose ambitious but controversial president was Thomas Watson Sr., who had previously been fired from his position as vice president of sales of the National Cash Register Company (NCR) of Dayton, Ohio. Within a few years Watson changed the name of his company to International Business Machines (IBM) and the rest is history.
His own fortune assured, Herman Hollerith remained as a consultant in engineering to IBM until 1921. He died in Washington, D.C., on November 17th, 1929, survived by his wife, the former Lucia Talcott, their six children and by more than 30 patents. Few if any of the later quantitative innovations in the marketing research field would have ever been possible without Herman Hollerith and his remarkable tabulating machine and keypunch card technology.