Jacquards' Web (23 page)

Hollerith also enjoyed watching the Jacquard looms in action in Meyer’s factories. Observing an expert work at a Jacquard hand-loom is an exhilarating experience; Hollerith was doubtless as fascinated by seeing Jacquard looms operating in New York as Babbage had been at witnessing a Jacquard loom working in Lyons.

160

The first Jacquard looms that wove information

The recollections of a professor at Cornell University, Dr Walter F. Willcox, who worked at the Census Office in
1900
, also show Hollerith’s fascination with the Jacquard loom. Willcox recalled conversations with the special agent Dr John Billings that had taken place about twenty years earlier. As Willcox remembered, Billings said Hollerith had discussed in detail with him how it should be possible to build a machine that could
automatically
count the findings of the Census and
record
the results on punched cards just as the Jacquard loom stored information on punched cards for weaving a particular pattern or design. In fact, Hollerith never pretended to have invented punched cards himself. He knew that the concept was Jacquard’s, not his. As his biographer Geoffrey Austrian points out, ‘Hollerith is often credited with the invention of the punched card, called the Hollerith card for many years and later more familiarly known as the IBM card. Yet, significantly, he never claimed it for himself. His basic patents always encompassed the use of punched cards
in combination with
his machines.’

[Austrian’s italics.]

But in any event, the entire concept of Hollerith’s machines is so closely related to the way the Jacquard loom stores the information necessary to complete a weaving that it is inconceivable Hollerith could have designed his machines without the inspiration of the Jacquard loom. It would certainly be gratifying if Hollerith had left behind any writing or notes that proved the matter conclusively, but in practice the circumstantial evidence for the link between the two technologies is overwhelming and decisive. The momentous idea that the key concept at the heart of the Jacquard loom could be transferred to weave information rather than fabric did not, after all, die with Charles Babbage.

Instead, like the smuggled silkworm eggs that brought the secret of silk to Europe, it merely went into a state of suspended animation. Almost twenty years later, in New York City, it woke up and started to breathe again.

And it has gone on breathing ever since.

161

Jacquard’s Web

During the early
1880
s, Herman Hollerith discussed in detail with his mentors Professor Trowbridge and Dr Billings the problem of finding a mechanical solution to the challenge of collecting and handling large amounts of Census information. Both special agents were fascinated by the idea, but ultimately it was Hollerith who embarked on practical experimentation directed at solving the problem.

By
1882
Hollerith had become convinced that the problem
could
be solved by mechanical means. He had also started thinking about what the machines that could solve it would look like and how they would work. In the meantime he had to feed himself, and unlike Babbage he had no family legacy to take care of this mundane but vital matter. This may have been an advantage to Hollerith; inventors driven by the necessity to earn money are often more focused and direct in their work and their approach than those for whom this is not so. Fortunately for Hollerith, in the same year
1882
, the Massachusetts Institute of Technology (MIT) offered him a post as an instructor in mechanical engineering. Hollerith was only twenty-two; the job offer was immensely flattering. He accepted it immediately.

Hollerith, all his life a powerhouse of energy and ideas, threw himself into his new job with devotion and gusto. Among the subjects he taught were the science of hydraulic motors, the design of machines, steam engineering, descriptive geometry, metallurgy, materials science, and even blacksmithing. Mechanical engineering was, at the time, the king of sciences. Mechanical engineers were seen, not unreasonably, as magicians whose tricks rarely went wrong. They built machines to carry out a myriad different functions, they constructed steam engines to power locomotives, ships, pumps, and factory machinery. Some mechanical engineers were already devoting themselves to meeting the challenge of creating horseless carriages. Armed with their specialized knowledge of metallurgy, hydraulics, power transmission, gearing ratios, and other lore, they were busily engaged on fashioning a new world all around them.

162

The first Jacquard looms that wove information

Indeed, mechanical engineering was developing so rapidly as a science that even Hollerith was taxed to the limit by the need to remain up-to-date with what was going on in the field. Computer engineers of today are likely to find that if they have a two-week holiday they may miss a crucial new development in computing.

Similarly, those wishing to keep abreast of mechanical engineering in the late nineteenth century had little choice but to keep working at the coalface where knowledge was being sledge-hammered out of the rock of ignorance. Yet despite the demands of his new job at MIT, Hollerith devoted his spare time to building various special machines for handling census returns.

Hollerith began his work by experimenting with methods he knew were already being used. These included strips of paper or individual cards on which details were noted by hand. But he was only too aware that the use of handwriting made the whole process extremely cumbersome. Besides, for Hollerith the purpose of a census was not merely to make a gigantic list of people’s names and addresses, but rather to analyse their lives and work in comprehensive detail. Soon he decided to continue his experiments using one of the very newest technologies available—electricity.

The first patent he applied for in
1884
covered the invention of a machine that represented information by a double row of holes punched across the width of a roll of narrow paper. Hollerith had decided to start his experiments by using paper tape because this was popular at the time as a means for storing information to be transmitted by means of the device known as the automatic telegraph. The automatic telegraph relayed information at a distance using electric current. In fact, the use of punched paper tape in the automatic telegraph also appears to have derived from the inspiration furnished by the Jacquard loom.

In Hollerith’s initial patent, the position of the holes indicated a person’s age, ethnic origin, whether a person was male or female and native or foreign born, or any other information that the Government wanted to record. Hollerith’s idea was that the 163

Jacquard’s Web

roll of perforated paper would be passed over a metal drum that advanced the paper and allowed it to interface with a counting device. Metal pins, forming a grid pattern in their layout exactly like the pins that pressed against the punched cards used in the Jacquard loom, were pushed against the drum by a spring. Each time they encountered a hole in the paper strip, they made contact with the drum, completing an electric circuit. The completion of the circuit activated an electromagnet whose action registered a ‘
1
’ on a counter provided for that hole.

The idea of having the presence or absence of a hole stand for a numerical quantity or a specific item of information would be absolutely fundamental to information processing in the years ahead. It was, again, exactly the idea at the heart of the Jacquard loom.

But Hollerith soon found that his concept of making use of a paper strip was not as practical as he had hoped. The paper tape turned out to be difficult to handle when it came to locating particular information. It also had an unfortunate tendency of snapping at critical moments. Abandoning the paper tape entirely, Hollerith adopted Jacquard’s idea of using punched cards.

The use of a punched card as the fundamental unit of his information processing system was the key to the whole success of the entirely new kind of machines that Hollerith now started to build. The punched card allowed units of information to be processed once, rearranged in new combinations, and processed once again until all the relevant information was obtained. In effect, the punched card became the world’s first-ever system for processing information using a standardized data input device.

Ultimately, the basic principle behind Jacquard’s loom, Babbage’s Analytical Engine, and Hollerith’s system of recording information about people was identical.

These pioneers all depended on using punched cards to record single, separate, units of information about something important they wanted to process or monitor, whether this was the specification of a picture to be woven into silk, the data and results of a 164

The first Jacquard looms that wove information

calculation, or the details of a particular person’s age, address, religion, country of birth, domicile, occupation, or—indeed—

anything else that was the subject of scrutiny.

One significant difference, however, between how Jacquard and Babbage had conceived of their inventions and how Hollerith visualized his was that Hollerith’s inventions, unlike the Jacquard loom and the Analytical Engine, were not stand-alone devices.

Instead, Hollerith’s vision encompassed an entire family of machines that would be deployed in unison. Today, this approach to using information technology is familiar to us all. Modern computers are routinely part of a system, linked to modems, printers, faxes, and—of course—each other. In fact, in the
1890
s, the idea of a family of machines being used to complete a task was familiar enough from the manufacturing industry; Hollerith simply borrowed this approach for his own work.

The first element in Hollerith’s system, understandably enough, was the
card-punch
: the device that actually punched the holes in the cards. In the early days Hollerith punched the holes himself using a stylus or other sharp object. But, he soon found this an unbearably tedious task, as well as hardly the most rewarding use of his time. Yet the job had to be completed with great application, because if the punched cards were to be an effective medium for storing information that could be processed by machines, the perforations needed to be in precisely the same place on each card to indicate the same answers to questions.

To carry out the task with the accuracy and speed required, Hollerith developed an ingenious machine based on the pantograph, a device invented in the early eighteenth century to copy a diagram on a larger scale. It consisted of a jointed adjustable parallelogram with tracing points at opposite corners. The picture on the following page shows how Hollerith’s pantographic punch worked. It allowed the operator to create a ‘mapping’ or symbolic relationship between the guideplate in the foreground, which showed the operator where the hole should be, and the 165

Jacquard’s Web

The pantograph punch.

actual card in the background. Provided that the operator indicated the right place on the guideplate, the pantograph punch would simultaneously punch the card in the right place. Using the pantograph punch, a clerk could punch many hundreds, or even thousands, of cards in a single day.

Once the punched cards had been prepared, the next stage was to count the occurrence of all the significant attributes of the person whose circumstances the card represented.

The ability of Hollerith’s system to facilitate the counting of different elements of data contained on the punched cards was central to the successful operation of his system of handling numbers and other kinds of information by special machinery.

The point is that the whole principle of Hollerith’s system was based on the notion of one card being used per person. An 166

The first Jacquard looms that wove information

The Hollerith counter and sorter.

operator could employ the system to retrieve specific information about their ages. The cards would all be individually pressed against a grid containing blunt-ended needles. This grid was very similar indeed to the card-reading device featured in the Jacquard loom. In Hollerith’s system, the needles were linked to dials which told the operator how many times the particular feature under examination had been triggered by the cards. The entire machine was known as the
counter
.