The data dump is a blessing for the caretakers of family trees, who can now mine that census for personal information about family members who passed on before their progeny could jot down key biographical facts. It’s also a long time coming; while aggregate statistics for cities or counties are published without restrictions as soon as they are available, specific records pertaining to individual citizens are sealed from the public, by law, for 72 years.
If that seems like a random amount of time to keep the decennial findings hidden, it kind of is. Sixty years ago, in an attempt to mollify both civil libertarians and statisticians who thought the value of the census was dependent upon confidentiality, Census Bureau Director Roy Peel and U.S. Archivist Wayne Grover wrote an informal rule (later codified by Congress) that forced the government to keep particulars under lock and key for seven decades. In 1952, female life expectancy in the States was 71.6 years. According to their logic, very few people would still be living who had participated in the census 72 years earlier, so any harm caused by the disclosure would be minimal. As it turns out, female life expectancy is now 79.5 years, and 21 million Americans alive in 1940 are still kicking today, a full 16 percent of those counted that year. Luckily, few have complained about a breach of privacy. Most, like 100-year-old Verla Morris, seem to enjoy the novelty of reading their name in America’s history book.
Back in the late-1800s, it took years to tabulate the census results at all. Bureaucrats didn’t put to bed the 1880 census until 1887, and they knew finishing the 1890 census by 1900, when Congress was constitutionally required to reapportion district boundaries, would be even tougher. Not only was the nation’s population expanding by about 25 percent each decade, but the Census Office added a series of new questions to the document, including queries about home ownership, war service, and race. Counting the data by hand, as they had done for a full century, wasn’t going to cut it.
Herman Hollerith knew just how inefficient the process was. Born to German immigrants in Buffalo, the eccentric Hollerith graduated in 1879 with an engineering degree from the Columbia University School of Mines and followed one of his professors into the Census Office, where he watched in horror as his new colleagues slogged through an endless pile of paper forms, one by one. Hollerith wanted desperately for the government to organize its records mechanically, thereby saving time and reducing errors. He just needed to figure out how best to do it.
Inspiration struck, as it so often does, on the train. As Hollerith recalls, he was taking a ride out from Washington when he watched a conductor use a punch card to certify a passenger’s ticket. That got him thinking: what if the government could transfer census questionnaires onto a punch card, with each hole representing a different data point (location, gender, occupation), and then feed the cards into an electrical machine that tallied the results? After five years of trial and error, the engineer finally figured out a design that worked. Using the same principles as a Jacquard loom, his prototype featured a series of tiny cups, all filled with mercury and connected to a wire nail. Each cup corresponded with a different hole on the punch card. When the card was inserted and the machine was set into motion, any punched hole would provide empty space in which the nail and mercury could interact like a circuit, thereby setting of an electrical charge. Those charges were sent to the machine’s dashboard, which contained a series of clock-like dials. All the census worker had to do was plug in a card, mark down which dials moved, take it out, and grab the next one.
Hollerith filed his first patent in 1884 and tested the gadget in Baltimore three years later. His old colleagues were impressed with the results and offered him a contract when they reopened for business in 1890. It was a profitable decision. Using the electrical invention, the Census Office was able to analyze more information in a shorter amount of time (five years) and at a discount to taxpayers (an estimated $5 million). “This apparatus works unerringly as the mills of the gods,” The Electrical Engineer wrote in November 1891, “but beats them hollow as to speed.”
Government officials may have been impressed with their new machine, but they were awfully cavalier with the documents it eventually tabulated. At the turn of the century, it was the job of individual agencies to maintain their own records, and some were more careful than others. Short on space in their vaults, archivists in the Commerce Department opted to stack the voluminous 1890 census neatly on pine shelves in their building’s basement. Few questioned the decision until January 10, 1921, when building fireman James Foster noticed smoke spewing through openings around some pipes that ran from the boiler room into the file room. Minutes later, another watchman upstairs smelled something burning in the men’s bathroom. Both made their way downstairs, where they ran right into an inferno. The pair pulled the house alarm, evacuated the office, and then watched as “five alarms quickly brought every piece of apparatus in downtown Washington to the scene” (New York Times, January 11, 1921). It took 20 hoses and two-and-a-half hours to extinguish the unfortunate blaze.
It was impossible to determine how long the fire had burned before anyone noticed, nor was it clear what set it off in the first place. (An errant cigarette is one potential culprit.) But the damage it caused was obvious. Kellee Blake, who wrote a big piece on the incident for Prologue, called it ”an archivist's nightmare.” One-quarter of the 1890 census burned instantly. Another 50 percent suffered heavy smoke and water damage. Census Bureau Clerk T. J. Fitzgerald told reporters the morning after that Hollerith’s data was "certain to be absolutely ruined” (Washington Post, January 11, 1921). And without modern preservation technology, the salvageable remains further deteriorated in the temporary storage space to which they were relocated. Today, only about 6,000 names from the almost 63 million census returns exist, a fact that frustrates genealogists to this day.
If there’s a silver lining to the story, it’s that the fire helped convince enough people in the capital that it would be useful to store important documents in a centralized and safe location. In 1926, Congress appropriated $1 million for an archival building, and eight years later, President Roosevelt signed a law establishing the National Archives as an independent agency. Hollerith, meanwhile, took the proceeds from his government contract and formed the Tabulating Machine Company, which would eventually change its name to the International Business Machines Corporation. He never became a rich man—the engineer did not get along with the company’s top salesman, Thomas Watson, and stepped aside from day-to-day operations in 1921—but his work revolutionized the field of information processing.
For more on that original contraption, be sure to read this article Hollerith wrote in 1890 describing its mechanics. The illustrations are particularly charming.