Past Times

Past Times: Counting machines improved the census

Punch cards were produced to represent each citizen of the United States in 1900.
Punch cards were produced to represent each citizen of the United States in 1900. U.S. Census Bureau

It’s hard to imagine the U.S. Census being completed without the help of computers. What’s surprising is how early the Census Bureau attempted to automate the process. By the late 1800s, it was already struggling to keep up with the flood of data being collected. The final tabulations for the 1880 census were not completed until 1887. Electric “counting machines” were leased for the 1890 census, but increasing costs led the Bureau to later develop its own enumerating machines. In 1900, Raleigh residents read about the marvels of “census electrical machines.”

Down at the Census office are a thousand or more electrical machines that can almost talk, and the labor they save is impossible to calculate. But for them it would require 10,000 clerks several years to tabulate and classify and compute the information sent in by the 50,000 enumerators regarding the 75,000,000 people who compose this great and glorious republic.

The first thing to be done when the returns of the enumerators are sent in upon large sheets of paper will be to transfer the information they contain to strong manilla cards about four by eight inches in size. The cards are impersonal but each bears a number and represents a citizen of the United States. Eighty million cards have been provided. Upon this card are 240 symbols, which represent every item of information that could possibly be reported by an enumerator under his instructions ….

This card is placed upon a machine that resembles one of those tracing instruments used for transferring maps, and the keyboard corresponds exactly to its lettering. The operator, glancing at the return of the enumerator, moves the key over the board and places it carefully upon the spot represented by the symbol wanted. This causes a little knife to pierce a hole in the card at that precise spot, and when the operation is completed the card resembles the perforated sheets that are used in orchestrions and automatic pianos.

When all the returns from a census district have been transferred in this manner from the enumerators’ sheets to the cards the latter are packed away in a tin box, properly marked, each box being large enough to hold 2,000 cards, and it can be easily stored away with other canned statistics, each box bearing a label indicating its contents.

I do not think I can describe the enumerating machine so that the reader can understand it. It is about the size of an upright piano. Upon the upright part is a set of dials, like steam registers, one for each of the items embraced in the population schedules. When the operator runs through this machine the manilla cards as above described and closes an electric circuit by moving a lever, a little rod runs through each hole in the card and causes the dial at the other end of a wire to register one. Therefore when the operator has completed the pile of cards and has run them all through the machine the information they contain has been accurately transferred to the several dials, and the operator can take the totals off on a slip of paper prepared for that purpose. The dials will show how many of the people represented by those 2,000 cards were born in Ireland and how many in Ohio; how many are married; how many are naturalized citizens, and all the other answers to the inquiries imposed by Congress upon enumerators.

The economy of this remarkable method was illustrated beyond a question in the last census, and can be appreciated when it is known that Alexander Maurice, one of the clerks in the last census, averaged 13,350 cards a day – while his highest record for any given day was 19,071 cards.

Anybody who will compare this work with the old-fashioned tally sheets, with closely packed lines of figures, which used to drive some of the census clerks blind and others crazy, can realize the value of the enumerating machine, which could not make an error under any circumstances. It is estimated that in the last census these machines saved not less than $600,000 in clerk hire. Both were invented by Herman Hollarith, a graduate of Columbia college, New York, who was a clerk in the census of 1880 and also in that of 1890, where he invented both devices with the assistance of Dr. Billings, now librarian of the United Libraries of New York; Henry Gannett of the geological survey, and others who took an interest in his work. He has since perfected the apparatus and is now preparing to place 1,000 punching machines and 250 enumerating machines in the census office. The N&O May 27, 1900

Read more stories from local and state history and send us your own stories on the blog Past Times,

Leonard: 919-829-4866 or