"The Perfectionists" by Simon Winchester, Harper, 2018.

I read the review, "Matters of Tolerance" by James Gleick, in the New York Review of October 25, 2018. The review is favourable; indeed so much so that I took the unprecedented step of actually buying the book.

A book review can do two distinct things: it can give the reader an idea of what the book is about and it can criticize it. Some reviews do both. Gleick's review is exceptional in refraining from criticism, or rather, giving an implicitly positive critique by writing a masterful précis. If you can somehow get hold of the review, then by all means read that. In the meantime I can do no better than to quote liberally from the review.

Engineers recognize a difference between precision and accuracy. The two qualities often go hand in hand, of course, but precision involves an ideal of meticulousness and consistency, while accuracy implies real-world truth. When a marksman fires at a target, if the bullets strike close together—clustered, rather than spread out—that is precise shooting. But the shots are only accurate to the extent that they hit close to the bull's eye.

Often we get spurious precision, spurious because it suggests accuracy that isn't there. For example π = 3.1456789 is precise, but not as accurate as its precision suggests whereas π = 3.141593 is. Because of lazy or ignorant programmers we are bombarded with numbers reported with more decimals than are relevant or warranted.

Precision implies not just a high degree of control over dimensions, but also the ability to make this reproducible. In late-eighteenth century Scotland, James Watt was designing a new engine to pump water by means of the power of steam. In England, John Wilkinson was improving the manufacture of cannons. They were cast as hollow tubes, and were prone to explode. Wilkinson invented a machine that took solid blocks of iron and bored cylindrical holes into them: straight and precise, one after another, each cannon identical to the next.

Meanwhile Watt found that his steam engine was too inefficient to be of any use. The problem was that the piston, which had to move within the cylinder, did not fit well enough to contain the steam within it. Enter Wilkinson, who wanted a steam engine. He saw the problem and had the solution ready-made. He could bore steam-engine cylinders from solid iron just as he had bored cannons, but on a larger scale. He made a gigantic boring tool of iron and achieved a cylinder, four feet in diameter, which as Watt later wrote, "does not err the thickness of an old shilling at any part."

After steam engines had faded away, the more general problem returned with later technologies: two things have be close to each other, yet move easily. There is a whole world of engineering in improving the fit and movement of pistons in the cylinders of internal combustion engines. Steam and gas turbines have to spin freely, yet stay close to their housing. Extremes of precision are needed to get read/write heads of computer disk drives to move at high speed and maintain an extremely small gap, with the constraint that the merest touch causes the dreaded, but fortunately rare, disk crash.

Watt's invention was a machine. Wilkinson's was a machine tool: a machine for making machines. The genius of machine tools—as opposed to mere machines—lies in their repeatability. Craftsmen can make tables or clocks exquisite and precise, but their precision was very much for the few. It was only when precision was created for the many that precision as a concept began to have the profound impact on society as a whole that it does today. That was John Wilkinson's achievement in 1776: the first construct possessed of a degree of real and reproducible mechanical precision—precision that was measurable, recordable, repeatable.

Winchester is a car nut. He abuses the theme of the book to treat the reader to his Rolls-Royce stories, the cars and the company. He gets back on track by contrasting these with those of Henry Ford, whose factory made more than a thousand cars for every Rolls-Royce. His invention was the assembly line. And though his cars were by comparison crude and cheap and unreliable, it was Ford, not Royce (the engineering partner), who demanded the utmost precision. The assembly line depended on replication, a flow of parts reliably the same, perfectly interchangeable. Rather than the cars, it was the process that was fine-tuned. The Model T went down in price from $ 850 in 1908 to $ 260 eighteen years later.

The car was the same, as were the materials; the difference lay in the increased efficiency of the process. A component that played an important role in this increase was the quintessence of precision engineering. That quintessence is embodied in a set of precisely-dimensioned, perfectly flat pieces of hardened steel known as gauge blocks, slip gauges, or Jo blocks, in the honour of their inventor Carl Edvard Johansson. These blocks are so flat that after putting one on top of the other, they cannot be pulled apart; they can only be separated by slipping sideways. Tantalizingly Winchester reports

... with 103 blocks of carefully selected sizes in three series it should be possible to take twenty thousand measurements in increments of one-thousandth of a millimeter by laying two or more blocks together.
My reaction here is: tell me more! What sizes were selected? What range of thicknesses was covered? Surely more than the 20 millimeters implied by the above numbers. With up to three blocks one can already make more than a hundred thousand combinations.

Winchester's book is fascinating: technical topics padded out with lots of supposedly colourful stories unconnected to technology. A more satisfying book would replace the latter material with precision-related topics. These could be organized by the dimensions of physics, of which the best-known ones are Length, Mass, and Time.

Length. In 1960 the metre was defined in terms of a certain number of wavelengths of a certain line in the spectrum of the rare gas krypton-86. It was the latest in a long quest for ever greater precision and reproducibility, perhaps beginning with the biblical cubit, said to be the length of Pharaoh's fore-arm.

Mass. Equally important for commerce was to have precisely defined and widely recognized standard. Of necessity it started in the period when the distinction between weight and mass was not recognized. My oldest readers learned in school that the standard of mass was that of a certain block of platinum kept at the International Bureau of Weights and Measures in France. Since then it has been redefined in terms of the speed of light, not something easy to measure precisely, one would think. But the idea is not to be dismissed out of hand, so that the redefinition represents the most solid basis. This is a story worth a place in the story of precision.

Time. While recognized as necessary, it was obvious that cubits, pounds, and stones (meaning those weighing fourteen pounds) were arbitrary. By contrast, for many centuries there was an obvious non-arbitrary unit of time: the Mean Solar Day, the time it takes for Earth to complete one rotation with respect to a hypothetical heavenly body called the Mean Sun. In theory it was known not to be perfect, but it was not before the 20th century that a more stable standard was found. Since then standard time has had to be adjusted by the occasional insertion of a leap second.

In summary, "The Perfectionists" is a good read and it identifies an aspect of technology that is new to the public. But it is to be hoped that it stimulates the appearance of a successor with a less idiosyncratic choice of topics.