Hollerith is not the father of the modern computer, but he is certainly the person responsible for the earliest form of automated data processing through his tabulating equipment and his punch cards. But more about that later, let’s start at the beginning.
Herman Hollerith was born on February 29, 1860, in Buffalo, New York, to German parents who had come to the United States in 1848. At school he was a bright pupil, but he had difficulty in spelling. As a result he often stayed away from school, as a consequence of which his parents removed him from school to be tutored privately by the family’s pastor. In 1879 he graduated from the Columbia University School of Mines with an Engineer of Mines degree. Shortly after graduation he joined the U.S. Census Bureau as a statistician working on the data of the 1880 census.
The U.S. Census Bureau
In 1790 the United States had a population of 3.8 million people. This had increased to 31.8 million by 1860. The Census Bureau conducted a census every ten years, the results of which were used, among other things, to determine the number of seats for each state in the U.S. House of Representatives. When it took the Census Bureau more then seven years to process the 1880 census data, they realized they had a problem on their hands with the next census. With the rapidly increasing population it was feared their next census’ processing might take more than ten years to complete. This would mean they might have to process two censuses simultaneously.
Hence it comes at no surprise that when Hollerith joined the Bureau he started to look at a method of mechanically processing this data. At the Census Bureau he met a woman, named Kate Sherman Billings, whom he courted for some time. Her father, Dr John Shaw Billings, who was the head of the Vital Statistics division at the Census Bureau, commented one day to Hollerith, “There ought to be a machine for doing the purely mechanical work of tabulating population and similar statistics.” Some say that Billings also suggested that the punched holes in cards that controlled the Jacquard loom could be used for that purpose. What is certain is that Billings gave Hollerith the inspiration to pursue a mechanical solution to the problem of processing the census data by automated means.
Hollerith’s punch card and tabulator
In 1882 Hollerith joined Massachusetts Institute of Technology (MIT) as a lecturer in mechanical engineering. During his stint at MIT he started experimenting on how to best record the data obtained in a census. During his train travels he had observed a train conductor punching holes in a passenger’s ticket to record details like place of departure, destination, and information about eye and hair colour, etc. The conductor used a simple handheld keypunch for this purpose. Initially Hollerith used a continuous strip of paper with holes in certain positions to indicate the information, e.g. one hole could indicate that the person was male, a hole in a different position that a person was female. By running the paper strip over a brass drum, brushes would detect the hole, complete an electric circuit and in doing so allowing a counter to register the electric current as a hit for the attribute that the hole denoted.
However, Hollerith soon realized that the continuous strip of paper had quite a few disadvantages: it was prone to tearing, while repair and restarting the process was cumbersome. Correcting an item was problematic, re-sorting items was virtually impossible, and finding a specific piece of data in a roll of paper meant the whole roll had to be read till the specific item was located. Looking for an alternative Hollerith turned to the same concept that Jacquard had used: punch cards. They had the advantage that correcting an item simply meant repunching a card to replace the incorrect one. Sorting was also easier and, since the cards ware made from stiff paper, damage was less likely to occur.
The cards became known as Hollerith cards, IBM cards, or simply punch cards. They measured 3.25” x 7.375” and were the size of the US one dollar bill, so Hollerith’s customers could use the same containers as the ones used for their currency to store the new punch cards [today’s one dollar bills are smaller than the 1880 ones as US banknotes were reduced in size by 20% in 1929]. One corner of the cards was cut off, so cards that were upside-down or back-to-front could easily be spotted. Initially the cards used round holes and had only 40 columns. In 1928 this would be increased and standardized to 80 columns with rectangular holes.
One could even send a punch card through the mail with a Valentine message in bad English
“Roses are red, violets are blue, my love for you has grew and grew.”
Hollerith had not realized that his system of punch holes was basically a binary system, the punched hole representing the binary “1” and the absence of the hole being the binary “0”. This would be a great advantage when punch card applications were transferred onto digital computers after World War II.
Translation: “1884 – Tabulating machine, the first data processing machine, built by
American engineer Herman Hollerith (1860-1929), used to supervise the 1890 census”
Hollerith’s machine to count and tabulate the punch cards was a sturdy desk-type construction, made of oak. The cards were read by a reading mechanism at the right and every time the reading mechanism encountered a hole in a specific position, a counter on the corresponding dial would be advanced by one. In total there were 40 dials (in four rows of ten), so 40 different criteria could be tabulated and counted at the same time. With later adaptations, Hollerith also managed to count combinations of holes, e.g. males between 50 and 60 years of age. Today that type of capability is quite common, but in those days the statisticians got very excited about this new facility.
Hollerith’s tabulator was first tested at the Office of Registration in Baltimore in 1884, where his system was used to record and count mortality statistics. It was so successful that soon Hollerith obtained similar contracts in New Jersey and New York City. He had realized that although he had developed his tabulator initially for the Census Bureau, having just one large customer every ten years would not pay his bills. So he urgently needed additional customers. In 1885 he installed his tabulators with the U.S. Navy which not only gave credibility to the usefulness of his machines but also provided him with the necessary financial resources.
The 1890 census
In 1888, the Census Bureau decided to hold a competition to find a more efficient method for processing its national census. Candidates would have to process data collected from four areas in St Louis and the time needed to come up with the results would determine the ultimate winner. Three people registered for the competition and started to work. The three candidates captured the data in 144.5 hours, 100.5 hours and 72.5 hours (the latter was Hollerith’s time). Then the data was processed by each candidate’s machine; this took 44.5 hours, 55.5 hours and 5.5 hours. The impressive result of 5.5 hours was achieved by Hollerith’s tabulator. Hollerith was the clear winner and he received the contract to process and tabulate the 1890 census data. The Census Bureau employed nearly 50,000 people to collect, capture (i.e. punch) the data, and tabulate its results. The 1870 census had only collected data on six criteria which had taken seven years to process The 1890 census covered 235 criteria, like sex, language spoken, number of children born, number of children still alive, etc. Despite this massive increase in data gathering, Hollerith’s tabulating system came up with the result only six weeks later: the United States had a population of 62,622,250. Although processing of the other data would take a further three years, Hollerith had managed to save the Bureau approximately $5,000,000 in manpower.
This very successful process also enabled Hollerith to base his dissertation on his tabulating system which he submitted to the Columbia University School of Mines and they awarded him his Ph.D. in 1890.
Following the success in the U.S., Hollerith managed to get his solution used in 1891 for the census in Norway, Canada, Austria, and later in England, Germany and Russia. Many other countries would soon follow using his punch cards and tabulating equipment to process their national census.
In Germany Hollerith operated under the name DEHOMAG (Deutsche Hollerith Maschinen Gesellschaft)
The Tabulating Machine Company
In 1896 Hollerith founded the Tabulating Machine Company to manufacture, market and sell his machines. He also insisted that his customers used his Hollerith cards in his tabulators as they had the necessary quality to limit the number of breakdowns of the tabulators, thereby avoiding costly maintenance. The cards were sold at a price between $0.80 and $1.00 per 1000 cards, which at a production cost of only $0.30 per 1000 gave his company a handsome profit margin. By leasing his machines, rather than outright selling them, combined with the continuous sales of his punch cards, he had managed to create a steady stream of revenue.
Private advertising on 1948 Remington Rand cover showing their variety in tabulating equipment.
Obviously competition soon arrived. Remington Rand would become Hollerith’s main competitor and became one of the largest suppliers of tabulating equipment. Commencing in the 1910s, he had an agreement with this competitor that they would not interfere with each other’s customers.
His first tabulators were all custom-built and even hard-wired for each application. Later he added functionality to his machines by incorporating a totaling function, an automatic card-feed mechanism, a sorting machine, a wiring-panel or plugboard (in 1906), so a machine’s functionality could be changed by just changing the wires in the plugboard. This was an early form of programming. While the first punch machine was a hand-held device, he soon devised the key punch which could be operated from a keyboard. In the 20th century all kind of additional pieces of equipment would be added: sorters, collators, interpreters (to print text on the punch cards “translating” the holes), and verifiers (to check that the data was captured correctly), to form an impressive array of equipment: the era of the tabulator. Soon his tabulators would be used by a multitude of companies for a large variety of applications: wages and payroll, bookkeeping, stock control, invoicing, insurance records, etc.
In 1911 Hollerith’s Tabulating Machine Company was merged with the Computing Scale Company of Dayton, Ohio, and the International Time Recording Company to become the Calculating-Tabulating-Recording Company (C-T-R). Hollerith stayed on at C-T-R in a role as consultant until his retirement in 1921. In 1924 C-T-R was renamed the International Business Machines Corporation (IBM).
Hollerith died from heart disease on November 17, 1929, in Washington D.C. and was survived by his wife Lucia Beverley Talcott and their six children. He was buried at the Oak Hill Cemetery in Georgetown, Washington D.C.
What makes Hollerith important?
Hollerith started the tabulator industry which reigned until the 1960s. The era of the punch card lasted even longer; as recently as 2000 the U.S. presidential election between George W. Bush and Al Gore was decided by hand-checking the punch cards used by the voters. Through his invention of the tabulator, which resulted in the Tabulating Machine Company, he stood at the birth of the giant IBM corporation, a company whose name is synonymous with computers.
What makes Hollerith even more important is that prior to his tabulators, the adding machines, calculators and cash registers could only process one transaction at a time. Hollerith was the first one to use a system that could process thousands of transactions (each one recorded in a punch card) in a single run. The concept of automated data processing had been born.
Through his tabulating equipment he accustomed many corporations to a systematic way of working. He continuously adapted his tabulators to satisfy the ever increasing number of requests from his customers: can the data be sorted, can the results be printed, can you add these two columns together and punch the result in the same card, etc. What Hollerith’s system obviously was missing was the concept of memory, but this was addressed when the digital computers arrived.
His customers got used to continuously thinking about their own processes and how they could improve their applications to improve their businesses. They got used to an application system using various pieces of equipment that could record large volumes of data, process it and print the results. Over the years their data increased in volume as did their need to manipulate this data, so they became prime candidates to accept computer technology when it became available in the post-World War II period. For decades computer applications would still be based on Hollerith’s ubiquitous punch card. Hollerith’s tabulating equipment was clearly the computer’s direct ancestor.
© Wobbe Vegter, 2007