Authors: James Essinger
The boys were raised by their mother. Her family had for many 152
A crisis with the American Census
years been locksmiths, a profession second in precision only to clock-making.
From the age of twelve Herman attended the College of the City of New York. In
1875
he transferred to the Columbia School of Mines. This seems an unpromising name for an academic establishment, but in fact the School had a reputation throughout the United States for fostering leading-edge technical work. It was a popular choice among talented young men intending to follow careers in science and mechanical engineering.
Hollerith’s academic career at the School of Mines was highly successful. He received perfect grades of
10
.
0
in descriptive geometry, graphics, surveying, and mechanical engineering. He also pursued a passionate interest in photography, carrying a camera almost everywhere. When he graduated in
1879
nobody doubted he would have a brilliant career.
One of the closest friendships Hollerith made at the School was with a Professor William Petit Trowbridge, head of the engineering department. Trowbridge was also a chief ‘special agent’ for the
1880
United States Census. This was planned to be the first US Census to focus as intensely on economic issues as on population data. The special agents were expert professionals charged with helping the US Government to make sense of the information the Census was expected to yield. Trowbridge had been given particular responsibility for analysing data relating to energy sources and mechanical engineering. It made sense to him to invite his most talented graduates to help him. As the most gifted of all, Hollerith was given the opportunity to play a particularly key role in the work.
Trowbridge decided to offer Hollerith the job of collecting statistics on steam and water power used in iron- and steel-making. Trowbridge had great faith in Hollerith’s powers. He introduced Hollerith to Dr John Billings, another special Census agent. Billings, as impressed with Hollerith as Trowbridge was, asked the young man to create a table of life expectancies for different age groups. Billings had particular responsibility for the 153
Jacquard’s Web
division of ‘Vital Statistics’ which, in those days, simply meant statistics related to living persons.
As with every project he ever undertook, Hollerith made every effort in completing these assignments to standards that were not only high but even obsessively so. Trowbridge was so pleased with Hollerith’s work that once the Census had taken place he allowed the young man the rare privilege of publishing a report on his findings under his own name.
Hollerith’s report investigated issues such as the relationship between fuel consumption and productivity and the number of workers employed in iron- and steel-making. Today, these hardly seem especially exciting issues to us, but back at the close of the nineteenth century, at a time when the United States was really starting to flex its industrial muscles, Hollerith’s work represented a valuable study. Furthermore, some of his findings were in fact quite momentous. He showed, for example, that while the use of water power had stayed almost constant over the past decade, there had been an increase of more than three hundred per cent in the application of steam-power to steel-making as a result of the new Bessemer and open-hearth steelworks.
This kind of information about industrial trends and major developments had never before been available in the US.
America had embarked on a remarkable curve of industrial development and population growth, but no tools had hitherto existed to gather the data that the government needed to monitor developments, control resources, and formulate policy. It was precisely this kind of data that the
1880
Census was designed to furnish.
The task of ploughing through the information generated by the
1880
Census was daunting. As for the US population, it was rising fast; the
1890
Census loomed on the horizon like a forbidding monster. It would have a terrifying appetite for research and statistics, but there were no tools available to feed that appetite apart from the hundreds of thousands of handwritten slips of paper that lay at the heart of the census-taking process.
154
A crisis with the American Census
The United States was a comparative latecomer in developing a requirement for large-scale information processing. In the
1830
s, when Britain, France, and Germany were making great strides in industrial development, the American economy was still largely an agricultural one. It was only after the end of the Civil War in
1865
that United States commercial organizations started to achieve significant growth. With this growth came industrialization. As the nineteenth century drew to a close, this became a scale of industrialization that no other country in the world could remotely match.
Yet the United States was actually helped in its drive towards industrialization in the second half of the nineteenth century by the comparative lack of existing industrial and technological infrastructure. It was not handicapped by an existing infrastructure which, if—say—it had industrialized at the same time as Britain, would by now have been eighty years old.
In particular, when the United States Government and individual corporations started to developed significant needs for information processing in the
1880
s and
1890
s, they weren’t hampered by the kind of manual information processing centres that abounded in Europe, where hundreds of clerks, working in cavern-like offices, kept voluminous ledger books by hand or made prodigious numbers of manual calculations. Instead, United States public and private sector organizations enjoyed the privilege of starting their information processing activities largely from scratch. They were perfectly positioned to make the most of any state-of-the-art information processing technology that was available.
This, combined with the United States having historically been blessed with many remarkably inventive technology pioneers, helps to explain how the US was able to establish and maintain a significant advantage in information processing over Europe at the end of the nineteenth century. This edge persisted throughout the twentieth century, and is even more pronounced today, in the twenty-first century, than it has ever been.
155
Jacquard’s Web
The prospect of the
1890
Census was placing the information processing requirements of the United States Government under great strain, but in fact the US Government’s need to gather and process information more efficiently had been an increasingly acute problem since the end of the Civil War. Before the War, the only data collection and collation organization of any size was the Bureau of the Census, based in Washington DC. The directive to conduct an annual census had been issued by Congress in
1790
in order to determine the ‘apportionment’ of members of the House of Representatives. The first ever United States Census, held the same year, estimated the young country’s population as
3
.
9
million. With a relatively small population to deal with, the information was processed by no more than a dozen clerks over several months.
The information gathered by the ten-yearly American Census continued to be handled in the same manual fashion throughout most of the nineteenth century. In
1840
, when the US population had risen to
17
.
1
million, there were twenty-eight clerks employed by the Bureau of the Census. By
1860
Congress had recognized that the Census could achieve far more than simply record population statistics. For this Census, the Bureau employed
184
clerks to deal with a census of
31
.
4
million people and to collate additional information about personal characteristics such as age, ethnic group, and occupation. For the
1870
Census there were
438
clerks to record data for
39
million people, and the final report amounted to
3473
pages.
By
1880
the US population had increased to some fifty million, which was primarily why the Census held in that year had placed such an immense strain on the Bureau’s information handling resources. To collate the results, a total of
1495
clerks were employed to run a manual system known as the ‘tally’.
After the Census forms had been completed and returned, each clerk had to transcribe the results from a particular district onto a grid containing a dozen columns and numerous rows. The columns and rows represented different parameters such as age, 156
A crisis with the American Census
sex, ethnic group, occupation, the name of the state where the person lived, and any other information that the Government wanted to record. These filled-in grids would then be analysed in an entirely visual manner by other clerks and eventually form the substance of a Census report.
Needless to say, this system was tedious, expensive, slow, and horribly susceptible to errors, much as errors—those
‘sunken rocks at sea’—had crept into the mathematical and scientific tables of the early nineteenth century.
The
1880
Census findings had taken seven years to collate in every level of detail required, although Hollerith himself was able to complete the statistics for his own work much sooner.
The
1890
Census would be even more massive than the
1880
one. The population of the US was shooting up; between
1880
and
1900
it increased to seventy-six million, a gain of more than fifty per cent in twenty years. Even without the benefit of this hindsight, it was obvious to everyone involved with the Census that it would not be long before collating all the results of a particular Census would still be going on when the
following
one took place. The consequences of this situation were not difficult to imagine: the US Government was in danger of getting itself into a position where it would
never
be able to catch up with itself in trying to monitor its rapidly increasing number of citizens and their activities.
Almost as bad was the fact that the need to handle everything manually placed overwhelming restrictions on the breadth and detail of the information gathered as well as severely limiting the extent to which the findings could be analysed and processed. At the very time when the US Government needed to understand changes in the economy and population of the nation better than ever before, it faced the prospect of having to
limit
the scope of its enquiries rather than
extend
them.
The need to find a better way to handle the findings of the US
Census was becoming desperate.
Fortunately, the Jacquard loom came to the rescue.
157
This page intentionally left blank
1 1 1 1 1 1 1 1 ❚ ❚ 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
2 2 2 2 2 2 2 2 2 2 2 2 2 ❚ 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
3 3 3 3 3 ❚ 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
4 4 4 4 4 4 ❚ 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
5 ❚ 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
6 6 6 6 6 6 6 6 6 6 6 ❚ 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 ❚ 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7
8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 ❚ 8 8 8 8 8 8 8 8 8 8 8 8
9 9 9 9 9 9 9 9 9 9 9 9 9 ❚ 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9 9
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ❚ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
The punched card was the key. It supplied the means by which units of information could be processed once, rearranged in new combinations, and processed again, until every bit of useful information was extracted. Hollerith had taken the crucial step in the development of his tabulating system.
Geoffrey D. Austrian,
Herman Hollerith, Forgotten Giant of Information Processing
,
1982
Herman Hollerith never publicly acknowledged the crucial role the Jacquard loom played in his work. Unlike Charles Babbage, Hollerith was not the kind of man to give credit where credit was due. As far as is known, he never made any public pronouncements on any subject. He never wrote an autobiography, nor did he leave a single word of writing that suggests any awareness on his part that he was carrying the sacred flame of technological destiny.
159
Jacquard’s Web
Nonetheless, we are justified in regarding the machines Hollerith created as a special kind of Jacquard loom that wove information rather than silk. In effect, Hollerith’s devices went a long way towards fulfilling Ada Lovelace’s inspired description of the Analytical Engine as a kind of Jacquard loom that ‘weaves algebraical patterns just as the Jacquard-loom weaves flowers and leaves’.
Hollerith’s tabulation machines were not as ambitious as the Analytical Engine, but they worked and they made commercial sense. Even more importantly, they paved the way for the creation of a new kind of machine—computers—which
would
offer the entire range of function and features that Babbage dreamt his Analytical Engine would provide.
The link between Jacquard’s work to develop a revolutionary loom and Hollerith’s creation of automatic accounting machines is clear and provable. It is a crucially important connection, and indicates that Herman Hollerith’s work is just as important a junction point between the Jacquard loom and the modern computer as Charles Babbage’s. Babbage laid the
conceptual
ground between the Jacquard loom and the computer, but it was Hollerith who made the connection a practical reality.
Hollerith knew all about the Jacquard loom. His brother-in-law, a businessman named Albert Meyer, was in the silk-weaving business in New York. Meyer had discussed the operation of the Jacquard loom in detail with Hollerith. Indeed, Meyer recognized Hollerith’s engineering talents, and tried to steer his brother-in-law towards the textile industry. He did not manage to do this, but the lengthy conversations Meyer and Hollerith had about weaving and textiles left Hollerith in no doubt that the Jacquard loom had pioneered a crucially important new way of storing information.