A fillet is a small radius provided at the junction of different shapes in a single part. For present-day motorcycling, the two outstanding examples of fillets are:

Under the head of a bolt, where the head joins the shank, where a crankpin emerges from the crankshaft cheeks or webs on either side of it, in a one-piece crankshaft. Because these are places where there is a sudden change of cross-section—from the larger bolt head to the smaller shank, or from the cylindrical crankpin to the much larger crank cheek—these are zones of stress concentration.

In 1974 the Yamaha TZ750A was introduced. Part of its teething process was occasional separation of a cylinder hold-down nut from its tubular shank. Kenny Roberts reckons such a breakage caused his bike to lose water in the Daytona that year, forcing him to ease up. Agostini won, Kenny was second.

I recently saw a video of how makers of high-strength fasteners prevent this kind of failure. They use a “deep rolling process” that rotates the bolt while a hard disc with a radiused edge is pressed hard against the fillet region under the bolt head. This rolling, by placing the fillet material in compression, means that no tension can occur in the fillet material until there is enough tension in the part to overcome this pre-compression. The idea of placing the surface of a part in compression as a means of increasing its fatigue life (today, typically by 2 to 4 times) is very old, and has surely been discovered independently many times.

In 1922 Bugatti were having crankshaft fillet cracking and failure in their race engine crankshafts. Specially-shaped fine hammers were prepared and workmen used them by hand—tap-tap—to place the crankpin fillets in compression. It was effective, which got people thinking. At other times, when shot-blasting was used to de-scale valve springs after heat-treatment, it was noted that springs so treated lasted usefully longer. This led, in the 1930s, to the process of shot-peening, using hard steel shot driven by air. The tiny overlapping dents they make in the surface force it into permanent compression.

Before shot-peening, aircraft engine connecting-rods had been laboriously polished to remove surface scratches and other irregularities that were known to act as nucleation sites for cracking. Shot-peening worked even better, even if the result was less beautiful.

After WW II, BSA were eager to market their own parallel-twin-powered motorcycles. This meant adopting cast or forged steel crankshafts in place of pressed-together 5-piece cranks like that of the revered BSA Gold Star single. General consternation was the result as test motorcycles powered by the new twin came back on the truck with broken cranks.

A consultant arrived to demonstrate how the auto industry was dealing with this problem. First he showed the failure mode. A brand-new crank was mounted with axis vertical on a shake-table. Such cranks consisted of a central flywheel with a crankpin on either side, in unit with smaller outer crank cheeks and then the mainshafts. With the shake-table’s output tubes’ emitters glowing red, the consultant made a frequency sweep. As he closed in on resonance (with the crank acting like a complex tuning-fork) a hum was heard from the vibrating crank.

As the assembled group of managers and engineers began to look at their watches, the consultant said, “This will only take five minutes or so…” Soon there was a click as the upper part of the crank broke off at a crankpin fillet.

As this was the early 1950s, and the German Hegenscheidt deep rolling process was not commercialized until 1957, BSA were probably advised to shot-peen the fillets. It made the parallel twins possible.

As TZ750s were made to turn almost a thousand rpm higher than intended in the early 1980s, and to make more power, cracking appeared in one of the inner flywheels of its two end-to-end twin cranks. I had no experience of shot-peening but sent the parts to be treated, with a sketch of the suspect region. Success!—cracking never again. I’d seen the eventual result—that flywheel disc would separate, becoming like a pair of 10,000 rpm steel brake shoes inside the thousand-dollar crankcase.

Another example of the increase of fatigue life by surface compression is rolled threads. The machine positions the blank between a pair of hardened grooved plates, which are then pressed together and slid past each other to forge the threads into the blank between them. Just takes a second or two.

In the early days of radial piston aircraft engines, failure of cylinder hold-down studs or bolts was constant. Threads cut with a die or on a lathe failed the earliest. Great hopes led to more accurate ground threads. Better, but not yet good. Rolled threads proved best. Today there is a process for rolling internal threads and at least some motorcycle manufacturers are using it.

That 750A also broke enough cylinder studs that I made a fixture for drilling out the broken stumps. In the textbook it says that for best fatigue life, the shank should be slightly smaller in diameter than the thread root diameter. Why so small? The idea was to eliminate the sudden change of section from the shank down to the thread root. Stuff in a book is just a rumor, right? The company promised warranty studs—new type—soon, but we had races coming up. I took a fresh set of original studs and gave them the book treatment, using a radius-nosed tool to reduce the shank diameter.

It wasn’t a rumor. No more breakages. I was delighted.

So much that we learn has been learned by other living things in a billion years of testing. Look at trees, whose branches join the trunk with proper fillets. The trunk gracefully flares at its base to become the root network.

The bones of birds are full of lessons—like a 787’s parts, they must be light enough to fly and heavy enough to arrive.