by Kilen Matthews
American inventor of a tabulating machine that was an important precursor of the electronic computer.
| Born: 29 Feb. 1860 in Buffalo, New York, USA
Died: 17 Nov 1929 in Washington D.C., USA |
Beginnings and Education
Herman Hollerith's parents immigrated, in 1848, to the United States
from Germany fleeing political disturbances there.
Hollerith was surely a bright child but struggled in his early schooling with difficulties learning spelling. There ensued a battle of will with his teacher resulting in young Herman avoiding school at all possible opportunities. Eventually, he was removed from school and received private tutoring at home from the family's Lutheran minister.
His higher education was a series of outstanding successes. Entering at the City College of New York in 1875, Hollerith then studied advanced engineering at the Columbia School of Mines in 1879. Hollerith earned honors and a distinction in his final examinations and so impressed one of his instructors, Prof. W P Trowbridge, that he asked Hollerith to become his assistant at Columbia University.
A Census Bureau in trouble...
When Trowbridge was appointed Chief Special Agent to the Census Bureau
he took Hollerith with him as a statistician. Among the areas Hollerith
was charged with was solving some of the problems of analysing the
large amounts of data collected and generated by the 1880 US census.
By the late 1880s, the U.S. Bureau of the Census was headed for serious trouble. The collating and analysis of the 1880 count of the then 50 million persons living in America had taken more than seven years to complete. With the population exploding to 60 million, it was estimated the 1890 census would require more than 10 years to finish, perhaps not being done in time for the next census to start!
As in many cases, the exact origin of the ideas for a great invention
is unclear and reports from different sources conflict.
Hollerith credits the idea of the census tabulating machine to a co-worker
at the U.S Census Bureau, Dr. John Shaw Billings.
Hollerith said, "One evening at Dr B's tea table he said to me 'There ought to be a machine for doing the purely mechanical work of tabulating population and similar statistics'."Billings also commented on the possibility of using a Jacquard loom process which involved the use of punched cards to specify intricate weaving patterns
In 1882 Hollerith was teaching mechanical engineering at the Massachusetts Institute of Technology. Working there, he experimented with the Jacquard loom to seeing if it could be used in census work. Finding the loom not really applicable to data storage and acquisition, Hollerith designed machines which used hole-punched paper tape to record and read data.
These machines worked but had several drawbacks which reduced their speed such as the need for the tape to stop to allow a metal pin to pass through the tape.
Hollerith concluded that punched cards would provide a better solution and an efficient way to store information. One day on a train, as he watched the conductor punch the tickets of riders with their destination Dr. Hollerith had a moment of inspiration: this was an easy way to punch information onto cards.
A Census Bureau saved
Hollerith's developed a series of devices using punched cards to record
and collate data. His system was first tried out for tabulating mortality
statistics in New Jersey and New York City in 1887 and it performed
well.
The U.S. Census Bureaus held trials of three candidate processing systems but Hollerith's was the clear winner. He quickly moved to have the card punching (Pratt and Whitney) and counting devices (Western Electric Company) manufactured.
All was ready by June 1990 and work began when the first census data
arrived at the Bureau in September of that year. The counting was completed
in six weeks, but the results were rechecked and held back until December
to reassure the public.
Instead of years, just three months were needed to complete the census
tabulation and determine that the population of the United States in 1890
was 62,622,250.
The 1890 census took advantage of the the speed and increased capabilities offered by the Hollerith cards and collected many additional data fields for the first time including items such as the number of children born in a family, the number of children still alive in a family, and the number of people who spoke English in the household.
The estimated saving through use of the Hollerith system were over 5 million dollars.
Hollerith's system enjoyed international success too, being used for
the 1891 censuses in Canada, Norway and Austria and for the 1911 UK census.
Hollerith Electric Tabulating System
How Hollerith Cards work
Numeric data is punched into cards represented by a single hole in
a designated area of the card. The letters of the alphabet are represented
by the combination of two punched holes on the same column.
The original version of the Hollerith punch card had room for 24 characters (digits or letters) with 12 rows for holes in each column. The final version had 80 columns, allowing up to 80 variables to be stored on each card.
Punched cards were read by the Hollerith machine one at a time very quickly. As each punched card was processed, a pin would fall through each card hole into a pan of mercury, closing an electrical circuit which registered the count on a meter.
In addition to the increase in processing speed provided the punched cards were sturdy and allowed retention of data and eliminated re-transcribing and computational errors which had taken so much effort in the past. This helped eliminate duplicate work and dramatically increased the productivity of the Census Bureau workers.
Into the business world
Hollerith did not enjoy teaching and sought a job in industry. In 1884
he took a position in the U.S. Patent Office in
Washington, D.C. It is not known if this was just good luck or a brilliant
far sighted career move by Hollerith. He gained important experience in
patents, applying for and receiving a patent for his card processing machine
in 1884. He would go on to receive more than 30 patents from the United
States and many additional foreign patents.
For the U.S. census of 1900, since Hollerith had a monopoly of punch card processing he set the price for use of his systems so high it was above what it would have cost to count the 1900 census data by hand. For the following count the Census Bureau decided to develop their own system and. The Census Bureau rushed to meet its deadline for starting the new census with their own machines, battling to get around Hollerith's patents, and they did so for the census beginning in 1910.
Hollerith's Tabulating Machine Company became the Computer Tabulating Recording Company in 1911 but soon lost the market for counting machines. Hollerith formed his a new company, TMC (Tabulator Machine Company) which produced card systems used in business accounting and railroad car inventory.
After several more mergers, TMC evolved into International Business Machines (IBM), the world's largest computer manufacturer, in 1924.
Honours and Acclaim
Although Hollerith had left the academic world, he continued to be
involved in it. He earned his Ph.D. in 1890 from the Columbia School of
Mines based on his work describing the details of his tabulating machines.
Hollerith became an internationally acclaimed speaker at places such as the Royal Statistical Society in London. He was awarded the prestigious Elliot Cresson Medal by the Franklin Institute of Philadelphia in 1890, the Gold Medal of the Paris Exposition and the Bronze Medal of the World's Fair in 1893.
Not all a bed of roses...
Although Hollerith made a very significant contribution to the development
of the modern electronic computer with his punched
card technology not all his ideas met with the same success. In the
1880s, while working on his punch card systems, he invented a new brake
system for trains. However his electrically actuated brake system lost
out to the Westinghouse steam actuated brake and was dropped.
Hollerith died of a heart attack in 1929, eight years after retiring.