The building of a time machine

Charles Babbage staked everything on his attempt to build the first computer, and lost. Overcoming similar problems, Doron Swade is about to meet the Babbage challenge 150 years on

A few years ago a smiling, bearded man in a waistcoat appeared at my office door, clutching three sheets of paper. I invited him in and listened to his tale. My visitor was Allan Bromley, from the Basser department of computer science at Sydney University. Bromley told me that since 1979 he had been visiting London for a few months each year to study the designs for a calculating engine of an English mathematician who lived 150 years ago.

The task was monumental. Charles Babbage's papers, held in the Science Museum library in London, make few concessions to would-be readers. Crucial sheets are undated. Comments and diagrams are cryptic and fragmentary. Complex mechanisms are depicted on large engineering drawings, but few have any written explanation. Over the years Bromley had succeeded in reconstructing the main tenets of Babbage's schemes and decoded a fair amount of detail. He was convinced that at least one of Babbage's engines would work. The document in his hand was a brief proposal addressed to the director of the Science Museum suggesting that the museum undertake to build a Babbage engine to original designs and that the project be completed by December 26 1991, the 200th anniversary of Babbage's birth. The proposal was dated May 20 1985. There was a little over six years to go.

The timing of Bromley's appearance was uncanny. I had just been appointed curator of computing at the Science Museum and was familiarising myself with the collections. I found tally sticks, quadrants, slide rules, arithmometers, comptometers, punched-card equipment and early computers. Among these riches one prized subset of objects stood out - the surviving relics of Babbage's efforts to construct his engines. These consisted of small working assemblies, experimental mechanisms and a portion completed in 1832 of his first Difference Engine. But no complete machines.

I began reading about Babbage in the books I had inherited with the job. The rudiments of his saga soon became clear. Babbage's first motive for devising the machines was to eliminate the risk of human error in the production of mathematical and navigational tables. The design for his Analytical Engine embodies most of the significant logical features found in the modern digital computer. The engine designs are acknowledged as ranking among the most startling intellectual achievements of the 19th century. But despite social privilege, political connections, independent wealth and decades of design and development, Babbage failed to construct a complete engine. It became evident that history celebrated Babbage as equally famous on two counts - inventing computers, and failing to build them. And almost every account gave the same reason for his lack of success - the limitations of 19th-century engineering.

In the end, faced with complex financial crises, an unreconciled breach with his engineer over disputed compensation, and the loss of credibility among his peers, Babbage abandoned his project.

The circumstances were complex. Babbage had a running feud with George Biddell Airy, Astronomer Royal for more than 40 years and de facto chief scientific adviser to government. Airy was consistently hostile to the utility of Babbage's engines. Babbage was convinced that Airy bore him a grudge that biased the astronomer's advocacy against the machines. History has taken Babbage's cue in portraying Airy as an unimaginative bureaucrat playing Salieri to Babbage's Mozart.

An unanswered question began to gnaw at my brain: could Babbage have built his engine and, if so, would it have worked? The next six years were dominated by an attempt to find an answer.

The engine Bromley proposed that we build was Babbage's Difference Engine No 2, designed between 1847 and 1849. Babbage made no attempt to construct this machine and the 20 design drawings survive intact partly because they were spared exposure to the hazards of workshop use.

The drawings depict a large machine 11 feet long, 7 feet high and 4 feet deep in places, consisting of 8,000 metal parts of bronze, cast iron and steel, weighing an estimated 5 tonnes. The engine calculates to 31 decimal places and was to be powered by a human operator turning a large crank handle located at one end. At the opposite end is a printing apparatus that records the results as an automatic outcome of the calculation process. The capacity to print results automatically is integral to the the machine.

It was too ambitious for us to attempt the whole engine in one go. No one had ever built a Babbage engine and there were inherent unknowns. While his drawings completely define the engine, they are not detailed enough to be used for modern manufacture. Essential information is missing: choice of materials, method of manufacture, precision and finish. The construction needed to be informed from a knowledge of 19th-century practice as well as of Babbage's design style. New drawings would be needed to specify the manufacturing detail for each of the 8,000 parts. As a staging post we did what Babbage did: built a small section of the engine as a test. The estimated time for completion was one year. It took four.

Our difficulties were uncannily similar to Babbage's. He had estimated that a few years were required to complete his first Difference Engine. Eleven years later he had completed only half the 25,000 parts needed. A decade later there was little more to show and the government formally abandoned the project.

There was no one simple explanation as to why in the mid-1980s the relatively small project to build the test piece took so long. On closer scrutiny the original drawings were found to contain layout errors, omissions and redundancies. Remedies had to be found consistent with contemporary practice.

National politics played its part in both attempts. Babbage worked on his engines during the most inventive as well as the most politically turbulent decades of the 19th century. There were squabbles over money, personal tragedy, professional vendettas and, in November 1842, a disastrous audience with the prime minister, Robert Peel.

The 20th-century sequel began during the cultural turbulence of Margaret Thatcher's Britain. National museums, allied to a public service culture, reeled under the shock of entrepreneurial conservatism and the requirement that they make money. Continuity was lost in the organisational upheaval and the Babbage project had to be restarted twice. By the time the test piece was complete there were only two years left out of the original six.

Reluctantly we abandoned the idea of building the printing apparatus for the bicentenary and lowered our target to the calculating section only. Babbage had done the same. The test piece was built in the Science Museum workshops but the full engine was way beyond our internal capacity. Parts manufacture had to be contracted out and funding became the issue. In the decade before 1832 Babbage was the beneficiary of massive sums from government, though the financial arrangements were messy. There was no budget, no completion date and the extent of Treasury commitment was never clear. By the time the engineers were paid off in the early 1830s the project had cost the public purse some Pounds 17,500 and was abandoned unfinished. This was a fortune by the standards of the day.

In 1989, the cost of manufacturing the parts of Difference Engine No 2 without the printer was estimated at Pounds 250,000. Sponsorship was the only option for a sum so large. IBM courted the project but insisted on sole rights whereby the company would fund the project in its entirety to the exclusion of other sponsors. This was risky. If IBM withdrew later the situation would be irrecoverable. But the appeal of being spared the slog of enticing multiple funders was too great to resist. Uneasily, Iwent with IBM.

In the late 1980s, the computing industry suffered a recession and IBM pulled out. Retrieving the situation was a theatre of extremes that included hijacking a trustee meeting attended by computer industry leaders at which I made an appeal to save the project.

The construction of the engine had a drama all its own: funding crises, a company going bust, impossible deadlines and incessant technical difficulties that arose as we carried the project into territory uncharted by Babbage. Throughout, there was the unremitting race against the clock.

The calculating section of the engine was completed in time for Babbage's 200th birthday on Boxing Day, 1991. But without the printer - which is integral to the engine - the machine was still incomplete.

The printing apparatus is more challenging than even the calculating section. From the standpoint of modern text processing, the mechanism offers a startling range of features. The apparatus is capable of programmable page formatting: variable line height, margin width, number of columns, insertion of blank lines and whether the results proceed down the page in columns or across the page in lines with automatic column or line wrap. All this is accomplished entirely by mechanical means to a design drafted 150 years ago. The printing apparatus is now being built in public view at the Science Museum with trials expected in the coming months. With the completion of the printing apparatus, we can finally see what no Victorian beheld - a complete working Babbage engine in all its splendour.

Babbage staked everything on his grand design - his hopes, ambitions and self-esteem. In his lifetime, the die fell badly and he never forgave his age for failing to acknowledge the wonder of his invention. He gambled so much and lost and was embittered by the experience. Perhaps with the completion of the engine now so close, this anguished episode in the prehistory of computing can finally be laid to rest.

Doron Swade is assistant director and head of collec-tions at London's Science Museum. The Cogwheel Brain: Charles Babbage and the Quest to Build the First Computer (Little, Brown) is published in April, Pounds 14.99.