Genealogists tend to focus on the individual census schedules as primary sources of family history information. But our use of these resources is quite incidental to the original purpose for which the census was taken—the apportionment of seats in the House of Representatives. Today, the government also uses the numbers from modern censuses to allocate over $100 billion in federal spending. But these purposes, the history behind the loss of the 1890 census, and invention of the census tabulating system that played such an integral part in data processing development are often overlooked by the family historian.

During the nineteenth century, industrial expansion and the flood of immigration created a need for technological advancements in what became known as data processing. Interestingly, the eleventh census of the United States—the "lost" census of 1890 bemoaned by researchers—stimulated this new technology. To tabulate census results, machines were invented that became the beginnings of modern information technology as we know it.

Herman Hollerith, the son of German immigrants, was a mechanical genius in an inventive age. Working on the manually-tallied 1880 census at the age of nineteen, Hollerith considered the enormous task of adding up the millions of census returns. In his travels by railroad (he had previously invented improvements for train braking systems), he had observed the use of a punch photograph. Conductors would punch notches in the edges of a passenger’s ticket which physically described the passenger, allowing only the described passenger to use that particular ticket. Hollerith envisioned a similar system by which the returns from the 1890 census could be punched onto paper cards. If he could automate the process for counting the punched cards, the census might be tallied in a far shorter time.

A pantograph punch would be used to punch the information onto the card. The operator would guide one end of a lever over a board showing the categories of information from the census (age, sex, place of birth, etc.) and would depress the human-readable end of the lever. The pantograph would simultaneously make the appropriate punch into a card. This would make the information from the census machine-readable. Each card was punched individually. Interestingly, the cards invented by Hollerith in the 1880s were used by the data processing industry into the 1970s.

The forward-thinking Hollerith also made his cards the same size as the "horse blanket" dollar bills of the day in order to take advantage of pre-existing cabinetry and other office equipment.

After the cards had been punched, they were individually inserted into the press of Hollerith’s tabulator machine. The press had yielding pins which, when they met the unpunched surface of the card, telescoped upwards. Where the pins were pressed through the holes, they connected with individual cups of mercury under the press and completed an electrical circuit. The circuit incremented a mechanical counter in the upper portion of the tabulator each time a hole in the card was detected by the pins.

To complete the process, Hollerith included an automatic sorting box attached to the tabulator. When the tabulator recorded a card with a particular characteristic, the lid of an associated compartment opened up. The operator of the tabulator placed the card into the correct box for further sorting. In this manner, additional tabulations such as "all persons living in Pennsylvania who are also steel workers" could be counted by running them through the tabulator again.

Hollerith’s tabulators had to be rewired each time a different set of characteristics were called for to be tabulated. This was the equivalent of re-programming the machines. For the 1890 census, Hollerith and an assistant soldered these wiring changes themselves.

Using Hollerith’s system, the rough population count of over 62 million individuals from the 1890 census returns was completed within six weeks. Completed statistical reviews from the 1890 data were fully published by 1892. Hollerith’s system had shaved years of effort off of the process and saved taxpayers approximately $5 million. At peak production, the Hollerith system processed census returns equal in height to the Washington Monument on a daily basis.

Herman Hollerith went on to win contracts abroad for census tabulation from Czarist Russia and the United Kingdom among others. The Hollerith system not only found a market in other corners of the world but was rapidly integrated for other applications. Once the railroad companies—the most complex business organizations of the day—introduced his system into their accounting departments, Hollerith’s success was assured. The company he eventually founded, International Business Machines, dominated the data processing industry until the 1980s. You can trace your own PC’s pedigree to Herman Hollerith’s automation of the 1890 census.

While working around the 1890 census gap using alternative sources such as state censuses and the special enumeration of Union Veterans and Widows which survived for some states, researchers may wonder what actually caused the loss of such an important research source? Arson? Accident? Neglect? Why did the 1890 census survive only in fragments of a meager 6,000 individuals out of the original count of over 62 million?

The answer to this question has been thoroughly researched and presented by Kellee Blake in her article "First In the Path of the Fireman—The Fate of the 1890 Population Census," which appeared in the Spring 1996 edition of Prologue: The Quarterly of the National Archives and Records Administration. See her article for full details.

In effect, Ms. Blake states that in March of 1896, the special schedules for the 1890 census, including those for mortality, poverty, and handicap status, were damaged by a fire and their remains were destroyed by order of the Department of the Interior. By 1921, the original and only copies of the 1890 census population schedules were stored in an unlocked file room in the basement of the Commerce Building, resting on pine wood shelving. A fire of unknown origin broke out in the basement on the evening of January 10. The Washington D.C. fire department contained the fire to the basement of the building with at least twenty fire hoses pouring water into the basement. In the aftermath of the fire, the Census Director estimated that twenty-five percent of the 1890 schedules had been destroyed, "with fifty percent of the remainder damaged by water, smoke, and fire."

By the end of January 1921, the remains of the 1890 schedules were moved out of the basement of the Commerce Building and placed in temporary storage. Between 1921 and 1932, the history of the census remnants is difficult to determine, but it appears that no salvage or restoration efforts occurred.

In keeping with standard federal recordkeeping procedure of the time, in December of 1932, the chief clerk of the Bureau of the Census sent the Librarian of Congress a list of papers to be destroyed. This list included the original 1890 census schedules. The Librarian was asked by the Bureau to identify any records which should be retained for historical purposes but the "Librarian identified no records as permanent and Congress authorized destruction on 21 February 1933." The remainder of the 1890 census was destroyed by government order by 1934 or 1935.

The 1890 census, whose enumeration was such a technical triumph of its day, was first damaged by fire and water then finally destroyed through disuse and indifference. But the Hollerith system used to tabulate that census set the standard for modern and efficient statistical enumeration for decades to come.

Mark Howells is a Certified Information Systems Auditor and a Certified Information Systems Security Professional. He volunteers on the Internet as the host of the Norfolk-L genealogy mailing list and is chairman of the Internet Branch of the Norfolk Family History Society.