Finding a solid solution

Elon Musk says “The best part is no part”. Elon Musk says a lot of things, but there are definitely times in history when this principle shines. The best example I’ve heard is the story of Charles-Édouard Guillaume, winner of the 1920 Nobel Prize for inventing the alloy Invar. Why is Invar Nobel worthy? Well, it has an extremely low thermal expansion coefficient. This may require some backstory.

In the 17^th century, pendulum clocks were an amazing new device. The original breakthrough came from Galileo, studying the motions of swinging objects. According to his student Vincenzo Viviani, a young Galileo in 1581 noticed that a chandelier seemed to swing at an incredibly consistent rate. Galileo’s later experiments with pendulums are more well documented, where he found that the period of a swinging pendulum stayed very consistent even as the arc of its swing dwindled over time. Christiaan Huygens used this principle to design the world’s first pendulum clock in 1656, and had it built by clockmaker Salomon Coster. The best previous clocks would lose or gain about 15 minutes per day. Huygens and Coster’s pendulum clock would lose or gain 15 seconds per day.¹

Clocks only improved from there. Robert Hooke designed a better escapement, making the pendulum swings smaller and thus even more consistent. This in turn let pendulums grow longer, which let them use less energy. Think of the “royal pendulum” – 0.994 meters long with one swing per second – found in the heart of grandfather clocks.

Pendulums rely on their exact length to maintain their steady beat. But the metal rods of clock pendulums expand with heat, lengthening during the summer and slowing clocks down by about a minute a week.² By the 1700s, clockmakers were getting so good at their art that they began to hit this level of precision. Unless thermal expansion could be solved, any further advances in making precision parts and mechanisms would be useless.

The first decent solution came in 1721, when London clockmaker George Graham replaced his pendulum weight with a glass vial of liquid mercury metal. As the rod of the pendulum lengthened downward with the summer heat, the mercury would expand too, rising in its vial and shifting the center of mass back into place.³

John Harrison offered another solution with the 1726 “gridiron pendulum”:

Notice that each side has 3 steel rods going down and 2 brass ones going up. Brass expands more with temperature than steel does. The downward-directed expansion of 3 steel rods can be counteracted by the upward-directed expansion of just 2 brass or zinc rods, leaving the weight at the end of the pendulum stable.

Harrison would eventually win the longitude prize for building the first accurate marine chronometers. They used springs instead of pendulums, but used the same principle of metals that expand at different rates to account for temperature changes.

These techniques worked well, achieving an accuracy of a few seconds per week, but they had some downsides. Mercury would take a long time to come to the same temperature as the pendulum’s rod, so the clock would run slow or fast if the temperature changed too suddenly. Gridiron pendulums were more responsive, but the fittings between the rods meant they didn’t expand continuously, but rather in sudden tiny jumps. By the mid 1800s, the highest precision instruments used mercury, while gridiron pendulums were used more widely when some precision was needed.

That’s where Invar comes in. Invar isn’t just a new mechanism, it sidesteps the problem entirely. Guillaume studied nickel steels, alloys of iron and nickel with just a little carbon. In 1895, he discovered that an alloy (specifically a “solid solution”) with 36% nickel dissolved in 64% iron basically doesn’t expand at all with temperature. In fact, if you quench it and cold work it enough, it can actually contract as temperature increases.⁴ You don’t need mercury or a gridiron with Invar – you just need a pendulum. Sometimes, the best part is no part.

The first pendulum clocks made with Invar, the Riefler regulator clocks, lost less than one second every few months. The best Invar pendulum clock produced at scale was the Shortt–Synchronome clock from 1921, which placed its primary pendulum in a vacuum chamber and used it to electromechanically drive a secondary pendulum in a more standard precision clock. These clocks were accurate to one second per decade.⁵ The original users of the Shortt-Synchronome clocks (like Naval observatories) didn’t even realize quite how accurate they were at first, because these were the first clocks more consistent than the Earth itself. Even the tiny shifts in gravity from the Sun and Moon passing overhead were detectable.

Eventually, even Invar was replaced. Piezoelectric quartz clocks took over in the 30’s, followed by atomic clocks in the 60’s. Today’s best atomic clocks have an accuracy equivalent to one second in hundreds of millions of years⁶ – though the physical clocks themselves won’t last that long. Physicists are now working on “nuclear clocks”, which in theory wouldn’t lose one second in hundreds of billions years. This kind of time precision lets us explore the fundamental physics of time, and could provide clues to about dark matter or changes in the constants of the universe.⁷

But despite all these improvements, Invar still holds a special place in the history of timekeeping. It has a lesson to teach us about the nature of engineering solutions, but its story goes even deeper than that. Once we had Invar, time was no longer read from the natural cycles of the planet, but dictated by the even more precise ticking of our pendulums. That’s why it’s the metal that remade time.

Coming soon: Some ballad meter poetry, to just amuse myself