The Dawn of Innovation (11 page)

The measuring procedure was straightforward. You fixed the standard bar in the grip with a precisely crafted “feeling piece,” an extremely thin sliver of steel. Whitworth stipulated a detailed protocol for inserting the feeling piece. You then slowly turned the division wheel, relaxing the pressure until the feeling piece moved, and you noted the mark on the division wheel. Repeat the process with a second bar, and the same feeling piece would expose any differences in length to a resolution of 1/1,000,000 of an inch. There was no ambiguity: when the feeling piece moved, you took the measurement and compared it to the same measurement for the standard bar. Whitworth's measuring machine was awarded the gold medal at the Great Crystal Palace Exhibition of 1851.
ⁿ

The Whitworth measuring machine was a pinnacle of nineteenth-century British precision engineering: a line of development rooted in the urgent search for precision in chronometers and astronomical apparatus, one that ran through the nearly perfect machine tools turned out by Henry Maudslay and the exquisite precision of the Nasmyth steam hammer. But the somewhat defensive comment by the Royal Astronomical Society that such precision was “not likely to affect any useful observation” was true, at least in a machine setting. Even today, ultraprecision computerized machine tools work at precisions of a few hundred thousandths of an inch.

Whitworth's measuring machine still pales beside the most audacious grasp at ultraprecise complexity. The ne plus ultra of regal overreaching was Charles Babbage's calculating engines.

If there were a hall of fame of intelligent people, Charles Babbage (1791–1871) would surely have his own plaque. Born into a well-to-do family, he spent most of his career in academia and for a dozen years held the Lucasian Chair of Mathematics at Cambridge University, a post graced by luminaries from Isaac Newton through Stephen Hawking.
²⁴

In Babbage's day, all calculation-intensive sciences like astronomy were dependent on thick volumes of standard tables—logarithms, sines, and other functions—each incorporating decades of laborious construction. As a newly minted mathematician, Babbage realized that even the best tables were riddled with errors. By comparing entries in different editions, it was clear that the primary problem lay in transcription and typesetting, not the original calculations. He therefore conceived of a machine, the Difference Engine,
^o
to infallibly compute and print such tables.

By 1822, Babbage had constructed a mechanical calculator that very rapidly executed standard arithmetic operations up to eight figures. He demonstrated it at the Treasury, and proposed that the government finance a much larger machine that would calculate and print any regularly sequenced table. With the Royal Society's backing, the following year, the Treasury provided a grant of £1,500. By his own account, Babbage undertook to complete the project “in two, or at most three, years,” with a commitment of his own funds of £1,500–£2,500.
²⁵

To execute the new machine, Babbage contracted with Joseph Clement, another former protégé of Maudslay. Clement is best-known for his large-scale metal planing machine, built sometime before 1832. The machine, with a planing bed that could hold work up to six feet square, moved on rollers so precise that “if you put a piece of paper under one of the rollers, it would stop all the rest.” For more than a decade, there was no other planer of its size in the world, so Clement, who charged by the square foot, made an excellent living keeping it operating almost around the clock. His partially completed Difference Engine has been called “the most refined and intricate piece of mechanism constructed up to that time.”
²⁶
Whitworth was one of the journeyman machinists employed by Clement on the engine.

Babbage had seriously underestimated the work involved and by 1827 had expended not only the grant amount, but beyond the estimated limit of his own funds. His version of what ensued shows him as a brilliant but almost comically contentious man. He duly applied for further assistance, but as a matter of right, on the grounds that the government had committed to see the project through, which seems a strained interpretation of the record.
^p
But he was backed by powerful friends and the Royal Society. Even the Duke of Wellington, then prime minister, became personally involved. Another £1,500 was awarded in early 1829, and yet another £3,000 later that same year, which Babbage at first refused to accept absent a firm open-ended financial commitment.
²⁷

Babbage had Clement construct a working segment of the Difference Engine in 1832 that, impressively, could produce a portion of the promised tables, and the government had already financed the construction of a small fireproof building to house it. By then, however, Babbage and Clement were constantly clashing over finances, and in 1833, Clement finally resigned, taking his machinery.

Amid the confusion, Babbage shifted his focus to an entirely new idea, which he called the “Analytical Engine.” It was nothing less than a prototype of the modern computer executed with purely mechanical parts. Delighted and inspired by the concept, Babbage worked on it alone for several years before admitting to the government that he was making a completely new start. Not surprisingly, ministers threw up their hands. Stubbornly, Babbage worked on the Analytical Engine for the next decade, using his own dwindling resources. To stretch his finances, he retired to a modest house with his own workshop and forge and a few assistants.

In about 1843, apparently realizing that the government was never going to finance his Analytical Engine, Babbage decided on a different
strategy: redesigning his original Difference Engine to take advantage of the new streamlined architecture of the Analytical Engine. He began to work on a “Difference Engine No. 2,” in 1846, producing a completely executed set of plans by 1849. This was by far the most practicable of his inventions: a limited-purpose Analytical Engine optimized to execute the mathematical tables proposed for the first Difference Engine. The proposed apparatus would have been much more efficient than the original, with only a third as many parts. Still, since the final product would have consisted of a wall of some 8,000 whirling parts, it would have been a substantial challenge.

To Babbage's shock and disappointment, the new government, which had come to office on an austerity platform, would not even entertain the idea of another prolonged engagement with Babbage: “Mr. Babbage's projects appear to be so indefinitely expensive, the ultimate success so problematical, and the ultimate expenditure so large and so utterly incapable of being calculated, that the government would not be justified in taking upon itself any further liability.”
²⁸
For Babbage, the cold rejection of his new plan was a serious blow that embittered much of his later life. He continued to produce books and articles on an array of subjects, wistfully tinkering with the design of the analytical engine until his death in 1871.

But the design is still an intellectual monument. Babbage's inspiration was from the Jacquard loom, which used punch cards to signify any pattern of threads whatsoever. Experts rendered patterns in sequences of punch cards. The textile mill then created additional sequences of cards to specify the thread colors for the pattern.

The Analytical Engine similarly had two main components, the mill and the store. A set of operation cards, similar to IBM punch cards, defined the sequence of operations to be performed by the mill, while a second set of variable cards summoned the sequence of data to be manipulated. Other sets of cards loaded the variables and constants into the store and defined where the results of an operation were to be stored. Output could either be printed or used to impress casts for printer's type.

The machine's registers—a small brass wheel for each digit—would accommodate fifty-digit numbers, which Babbage thought sufficient for science, along with 1,000 stored constants. Sequences of cards were placed in a card reader and called on in the proper order, and all standard operations would be maintained in libraries of cards. (Babbage also worked out a provision for an unlimited number of if-then, “looped” instructions.) Gear layouts and speeds were such that the engine should have been able to perform about sixty operations a minute.
²⁹

Babbage topped it off with another invention, his “Mechanical Notation”—a language for specifying rigorously any machine whatsoever (or, in Babbage's mind, almost anything at all, including physiology, factory organization, or war planning).
³⁰
It is a detailed coding system for machine parts, actions, and motions that, as he explained to the Royal Society in 1826, reflected the actual machine with such precision “that at any moment of time in the course of the cycle of operations of any machine, we may know the state of motion or rest of any particular part.”
³¹

Babbage told the Society that he had been compelled to develop the notation as he grappled with the complex gear sequences of the Difference Engine. Once a machine was completely rendered in notation, one could readily see whether the design would work or not, and how to simplify and improve it, without the expense of building a model. Once he had described the Difference Engine in notation, Babbage claimed, he could spot design issues much faster than the artisans. The Society presentation was illustrated by a notational description of an eight-day clock. Comprising four oversized pages of dense columns of cryptic markings, it is a beautiful and astonishing production, rather like a Japanese archival scroll. The drawings, notebooks, and other documentation for the nearly completed design of the Analytical Engine comprise nearly 7,000 pages, of which 2,200 are purely notation.
³²

Babbage thought the notation was “one of the most important contributions I have made to human knowledge. It has placed the construction of machinery in the ranks of demonstrative science. The day will arrive when no school of mechanical drawing will be thought complete without teaching it.”
³³
Once again, however, Babbage was pointing directly to developments
still far in the future. It was only with the onset of the postwar Computer Age that technologists began to execute their chip and other hardware designs in software so they could be tested and exercised without the expense of building physical components.

The tantalizing historical question for Babbage admirers was always whether his machines would actually have worked. That was finally answered by an extraordinary project at the London Science Museum that built a working model of the Difference Engine No. 2 (DE2 hereafter). And yes, it actually worked.

The DE2 was the redesigned Difference Engine No. 1 using the new architecture Babbage had created for his Analytical Engine. It was the smallest of Babbage's engines and the only one with a completed design, which happened to be stored, along with its Mechanical Notation, in the museum's archives. Alan Bromley, a computer scientist at the University of Sydney, made a study of the plans over a number of years and convinced himself, and then the museum staff and directors, that it was feasible to build and would probably work as Babbage claimed.
³⁴

The project got underway in 1985, under the direction of Doron Swade, then the Science Museum's head of collections and an expert on Babbage and the history of computing, working on a tight six-year schedule, timed to the Babbage bicentennial in 1991. The final construct was seven feet high, eleven feet across, and eighteen inches deep. It had eight tall columns, each with thirty-one figure wheels, all of it operated by a hand crank. A few drawings had to be reconstructed, and Babbage's designs also included a small number of mistakes—this was, after all, a purely paper design, with no testing of individual modules—but they were usually obvious and readily fixed without violating the integrity of the design.

Much more impressive are the mistakes Babbage didn't make. For example, one of Swade's engineers looked at the design and said it would be impossible to run from a hand crank—the combined resistance from 4,000 gears meshing was just too great. The team went ahead anyway, and of course the engineer was proved right. Then they noticed a mysterious spring mechanism that Babbage had enclosed within the engine frame. Since no one knew its function, it had not yet been installed. And, yes, the purpose of the spring arrangement was to ease the frictional pressures so the hand crank worked as envisaged—an altogether astonishing degree of foresight for a paper design.

 

A modern realization of Charles Babbage's Difference Engine No. 2
, a distant prototype for the modern computer. It was constructed at the London Science Museum over a six-year period, and completed in time for the Babbage centennial in 1991. The vertical columns mostly contain brass wheels—some 4,000 in total—each of which represents a digit or part of an action. The construction proved that Babbage's paper designs worked the way he said they would. The modern reconstruction used current machine-tool and dimensioning technology. Whether the machine could have been built with 1830s technology is still an open question.