Page 1 of 3
Computers compute and to most people that means doing calculations. The early history of computing is full of tales of scientists and engineers being driven to invent computing machines from scratch because of the mind-numbing tedium of having to perform repetitive numerical calculations. Of course it is generally reckoned that these early scientific computers grew into the data processing machines that the business community needed without any former effort on their part to automate their data processing.
This isn't true. The scientists needed machines that could calculate but the businessmen needed machines that could process data and they needed them before the scientists even noticed that they had a little difficulty with numbers.
In the field of business computing one man can be credited with inventing automatic data processing - an achievement every bit as impressive as building the ENIAC or the MARK I - but these days his name is hardly known. You might even call Herman Hollerith, the forgotten giant of computing.
(February 29, 1848 - November 17, 1929)
Herman Hollerith was born in Buffalo New York. His father had fled Germany in 1848 and died when Herman was only seven. He was a very practical child and had a natural engineering talent. At fifteen he entered the Columbia School of Mines. His summer holiday work was in the coal mines around Michigan. In 1879 he graduated with a degree in mining engineering. It looked as if his future was settled but the US was one year away from the ten-yearly census and there were problems that needed a mechanical mind to solve.
The US Census
The need for a US census grew as the size of the population increased and of course the difficulty in conducting the census similarly increased. The first census began in August 1790, took nine months' of data collection and took about a year to count. Every ten years after that the US repeated the census but the size of the population was growing at 35% per annum and by 1860 there were over 31 million people to count.
Over the years the census had also grown in complexity - so much so that in 1850 the congress passed a law to limit it to no more than 100 questions! It was beginning to look as it might take more than 10 years to process the data and what good would 10 year-old statistics have been. Modern government needed up-to-date information to plan services and predict taxation.
It is difficult to understand the nature of the census problem from our perspective. We naturally think of using computers to handle large quantities of data but in 1850 there was no electronics and the census data had to be processed by hand. Clerks had to move huge mountains of paper and count the responses by hand. This not only resulted in long processing times, it also limited the sort of questions that could be asked. You could obtain simple frequency counts but building tables of categories, i.e. a cross tabulation, was even more time consuming.
So difficult was the task that even very simple mechanical aids paid off. For example, in the 1870 census Charles Seaton built a machine that provided a paper roll that the clerks could write on. Helping to keep things neat did speed up the counting - but something even more radical was needed.
After graduating Hollerith was offered a job helping one of his professors, William Trowbridge, compile a report for the census on the use of steam and power. Hollerith had very little work to do and was able to engage in the society life of Washington. His only hobby was photography but like many a photographer he had a dislike of having his photo taken!
Hollerith dated Kate Sherman Billings, an unremarkable event in itself except for the fact that her father was in charge of a large section of the census - vital statistics. Hollerith tried to impress Kate by trying to buy every one of the tickets for a lottery at the party they were both at. Unfortunately he missed just one ticket - the winning ticket! At the buffet he chose the chicken salad and this gave Kate the opportunity to invite him home where her mother made a great chicken salad!
At the Billings' home Hollerith got into a deep conversation with Kate's father about the census problem. Billings was sure that a mechanical method of counting could be built but he was a physician and not an engineer. Later Hollerith reported their conversation:
"He said to me that there ought to be a machine for doing the purely mechanical work of tabulating population statistics... He thought of using cards with the description of the individual shown by notches in the edge... After studying the problem I went back to Dr Billings and said that I could work out a solution and asked if he would go in with me. The Doctor said he was not interested any further than to see some solution worked out."