Columbia University Computing History   
Translations  (see below for credits):
Finnish | Suomi French | Français Georgian | ქართული German | Deutsch Italian | Italiano Norwegian | Norsk Polish | Polski Romanian | Română Russian | Русский Spanish | Español Ukrainian | Українська

Hollerith 1890 Census Tabulator and the evolution of the IBM punch card

Herman Hollerith was a Columbia graduate and later received a Columbia PhD for his 1890 census work.

Hollerith 1890 Census tabulator
Photo: IBM.
The image shows Herman Hollerith's 1890 tabulating machine (image from IBM; CLICK HERE for a color photo). The results of a tabulation are displayed on the clock-like dials. A sorter is on the right. On the tabletop below the dials are a Pantographic card punch (explained below) on the left and the card reading station ("press") on the right, in which metal pins pass through the holes, making contact with little wells of mercury, completing an electrical circuit. All of these devices are fed manually, one card at a time, but the tabulator and sorter are electrically coupled.

1890 Census process
Images: [103]; CLICK to enlarge.

Left to right: The circuit-closing press ("card reader"); diagram of press; hand insertion of card into a sorter compartment that opened automatically based on the values punched into the card; tallying the day's results. "Each completed circuit caused an electromagnet to advance a counting dial by one number. The tabulator's 40 dials allowed the answers to several questions to be counted simultaneously. At the end of the day, the total on each dial was recorded by hand and the dial set back to zero [103]."

1890 Census tabulator
Herman Hollerith at tabulator (1894) - click to enlarge
This apparatus works unerringly as the mills of the gods, but beats them hollow as to speed.
 –The Electrical Engineer,   11 Nov 1891.   

The 1890 tabulator was capable only of counting. Subsequent models, developed by Hollerith himself, were also able to add, thus broadening their scope to accounting, warehousing, and shipping applications. Between 1902 and 1905, Hollerith also developed an automatic card feed and a method for reading cards in motion and settled on a standard card format. In 1928, IBM produced its first tabulator (the Type IV) with both addition and subtraction capability.

1890 Census keypunch
Photo: (from the 1920 census): [44].
Census tabulating 1890
Tabulating the 1890 census; Image: Scientific American

Left: The Pantographic card punch in operation. The operator holds the stylus over the template. The card is in the punch station above the template. Right: After the cards are punched, they are fed into the Hollerith tabulators.

Pantographic punch-card template

Above: A punch-card template from a Pantographic punch used the 1890 US census (image: US Library of Congress). Notice there are 12 rows (as in modern punch cards), of which only the bottom 10 were used, and only 20 columns; the curved shape is due to the Pantographic mechanism, an early ergonomic device allowing operators to punch 500 cards per day with good accuracy and minimal physical strain (compared to the handheld "train conductor" punches used in previous trials, which could cause near paralysis with prolonged use -- carpal tunnel syndrome did not start with PCs! -- and with which correct placement of holes was problematic). The Pantographic punch operator positioned a stylus over the desired hole in the template and pressed it to punch a hole in the corresponding position of the rectangular card. The template has areas marked off for various demographic categories.

1890 Census card layout

Above: The "reading board" for a punched card from the 1890 census (the cards themselves were blank; this is like the "decoder ring" for the holes, which itself needs a second "decoder ring" to decipher the alphanumeric codes). Cards had one one corner cut diagonally to protect against upside down and/or backwards cards that might not otherwise be detected and the reading board had the same cut for obvious reasons. The card measures 3.25 by 7.375 inches... (image and dimensions from Aspray, Computing Before Computers[69]).

On 16 August 2023, I received the following email from Keith Elkin:

Aurora Tucker, the collections Manager at the Computer History Museum in Mountain View, California went out of her way and took measurements of their 1890 Hollerith pantograph. I thought I should pass this on to ensure that the actual values become part of the historical record. This gives the possible size of the Hollerith punch card in 1890.

(Aurora Tucker) I am pleased to share the following measurements:

  • The width (widest card dimension) is 6 5/8 inches, which can be expanded by a maximum addition of 3/8 inches if the spring arm is fully extended.
  • The depth (shorter card dimension) is a static 4 ¼ inches.
... the same size as the 1887 US paper currency because Hollerith used Treasury Department containers as card boxes (pictures not actual size, but all to the same scale):

US ten dollar bill 1901
United States ten dollar bill 1901

US banknotes were reduced in size by 20% to their present dimensions in 1928:

US ten dollar bill 19281
United States ten dollar bill 1928

Using round holes, the card column density eventually reached 45 prior to the 1928 80-column rectangular-punch standard. Here is the modern (or at least, final) standard, corner-cut, 80-column general-purpose IBM punch card, introduced in 1928, and popularly known as the "IBM card":

1928 Standard IBM punch card
80-column IBM punch card, 1928 and later - click image to enlarge
Holes in the 80-column card are rectangular, rather than round as in earlier models. The bottom ten rows are labeled with digits; the top two rows are unlabeled and are used in an alphanumeric character code first standardized (by IBM) in 1931 as BCDIC, a 40-character set that included digits, uppercase A-Z, space, minus sign, asterisk, and ampersand [52], eventually expanded to a large family of 256-character Extended BCDIC (EBCDIC) codes, IBM's Country Extended Code Pages. This type of card was a mainstay of data processing and computing from 1928 through the 1980s and was still in use in voting machines through the 2000 USA presidential election, in which they were discredited when the number of swing ballots might have been less than the number questionably-punched cards and therefore contested ballots (we'll never know, since they weren't counted). Although the poorly punched cards were due primarily to unmaintained machines (many of them more than 40 years old), most localities resolved to replace punched-card technology with something more modern, like optical scanners. Whether the new technology is more reliable, accurate, cost effective, durable, and resistent to fraud and tampering is a concern, but there is apparently no going back.

  1. The Hollerith Machine, United States Census Bureau website, accessed 16 April 2021.
  2. 1890 Census: The Meaning of the Hollerith Card Codes,, accessed 16 April 2021.
  3. "The Hollerith Method of Statistical Tabulation", Frank Leslie's Illustrated Newspaper, October 12, 1889, p.182.
  4. Scientific American, Vol.63, No.9, August 30, 1890.
  5. Martin, T.C., "Counting a Nation by Electricity", The Electrical Engineer, New York, Vol.12, November 11, 1891, pp.521-530.
  6. "Hollerith's Electric Tabulating Machine", Railroad Gazette, 19 April 1895.
  7. Austrian, Geoffrey, Herman Hollerith: Forgotten Giant of Information Processing, Columbia University Press (1982) [44].
  8. Bashe, Charles J.; Lyle R. Johnson; John H. Palmer; Emerson W. Pugh, IBM's Early Computers, MIT Press (1985) [4]
  9. Eames, Charles and Ray, A Computer Perspective: Background to the Computer Age, Harvard University Press. First Edition 1973; Second Edition 1990 [103].
  10. Knuth, Donald, The Art of Computer Programming, Vol.3 "Sorting and Searching", Addison-Wesley (1973); Section 5.5, pp.382-384 [104].
  11. Pugh, Emerson W., Building IBM: Shaping an Industry and its Technology, The MIT Press (1995) [40]
  12. Truesdell, Leon E., The Development of Punch Card Tabulation in the Bureau of the Census 1890-1940, US Government Printing Office, Washington DC (1965).
  13. Destruction of the 1890 Census,, accessed 20 October 2019. The story of how the 1890 records (including forms and cards) were destroyed in 1932, before the National Archive was established.
  14. The Punched Card Tabulator, IBM 100: "Herman Hollerith's first tabulating machines opened the world's eyes to the very idea of data processing. Along the way, the machines also laid the foundation for IBM..."

Translations of this page courtesy of...

Language Link Date Translator Organization
Finnish Suomi 2023/08/31 Kerstin Schmidt
French Français 2023/08/25 Kerstin Schmidt
Georgian ქართული 2019/08/20 Ana Mirilashvili LPA Code Professional Scientific Translations
Italian Italiano 2023/08/31 Kerstin Schmidt
German Deutsch 2023/08/25 Kerstin Schmidt
Norwegian Norsk (bokmål) 2022/08/11 Rune Bildeler på nett
Polish Polski 2023/08/25 Kerstin Schmidt
Romanian Română 2023/08/17 Vladyslav Byshuk | Владислав Бишук
Russian Русский 2023/08/17 Vladyslav Byshuk | Владислав Бишук
Spanish Español 2023/08/31 Kerstin Schmidt
Ukrainian Українська 2023/08/17 Vladyslav Byshuk | Владислав Бишук
Columbia University Computing History Frank da Cruz / [email protected] This page created: January 2001 Last update: 4 September 2023