The engine everybody bought and almost nobody made stick

You Spin Me Round

Say “rotary” and people say “Mazda.” But the Wankel was tried in Citroëns, a Mercedes prototype, a Rolls-Royce diesel, a Ford Mustang, snowmobiles, lawnmowers, outboards — and, best of all, motorcycles.

By Douglas James · 30 minute read

Here is the part of the story that got lost. For about fifteen years, a large slice of the industrialized world was not merely curious about the rotary engine. It was hedging against the possibility that the rotary would become the default, and that anyone still building piston engines would wake up obsolete.

Roughly 100 companies stood in line for a license. Aircraft manufacturers, truck builders, luxury carmakers, motorcycle firms, outboard-motor companies. Nearly all of them failed. The one that succeeded was a Japanese firm that had started out making cork.

That is the famous half. This is the other half.

It started on a moped

Before the rotary was an engine, it was a supercharger.

Felix Wankel went to work for NSU in 1951, and the job he was hired for was not the one he is remembered for: he was brought in to build a rotary valve motorcycle engine. NSU itself was not a car company at all. It had begun life as the Neckarsulm knitting machine works, rebuilt after the war on the back of a cheap 50cc moped called the Quickly, and was run by a former Opel manager who had been made to start over as a salesman because of what he had done during the war.

What Wankel actually built first, in the early fifties, was a rotary compressor: a nearly triangular inner rotor turning inside an outer rotor, both spinning around a fixed shaft. It was designed to supercharge motorcycles, and it worked extraordinarily well. NSU put one on a 500cc twin and produced 110 horsepower from it. In 1954 they fitted one to the 50cc Quickly and set a speed record of 120 mph on the Bonneville Salt Flats.

A 50cc moped at 120 mph. Somewhere in Neckarsulm there was presumably a warranty department that was never told. The rotary’s first real-world triumph happened on two wheels, years before it went into anything with a roof.

What everyone thought they were buying

When Wankel’s engine started appearing in technical journals in 1960, the specification read like a list of solved problems. It was about half the physical size of a piston engine of comparable displacement. It had no valves, no camshafts, no valve springs — eliminating the valve drive alone saved hundreds of parts per engine. Nothing inside it stopped, reversed, or changed direction, so it ran almost free of vibration. And at an ordinary 8.5:1 compression ratio it would run happily on low-grade fuel, which mattered enormously in places where high-octane gasoline was expensive or simply unavailable.

Designers looked at it and saw ultra-low hood lines and mid-engine layouts. Small-engine manufacturers looked at it and saw a replacement for the two-stroke in chain saws, lawn equipment and outboard motors. Companies with heavy-industry ambitions looked at it and started talking about stacking rotors in modules to build engines of any size they liked.

The engine was small enough to fit almost anywhere, and that turned out to be the problem as much as the promise. Everyone could imagine a use. Almost nobody could make one last.

The Americans got there first

The first company in the world to buy a Wankel license was not a carmaker. It was Curtiss-Wright — the direct corporate descendant of the company the Wright brothers founded to sell their flying machines, merged in the twenties with the firm of Glenn Curtiss, the man who sold the United States military its first airplanes. By the fifties it was a defense contractor whose aircraft engines were being made obsolete by the turbojet, run by a president named Roy Hurley who was hunting for the next thing.

The next thing arrived as a model carried home from Germany. A Studebaker representative visiting Germany on fuel-injection business heard about NSU’s new engine, went to look, and was handed a model to take home. It reached Hurley, who handed it to Max Bentele — a German-born turbine engineer who had spent the war on jet blade design and been persuaded, some years afterward, to move to the United States — and told him to study it over the weekend.

Bentele’s verdict was that the geometry was elegant and the sealing would be brutal. He advised Hurley to buy a $50,000 option (about $580,000 in 2026) rather than a full license, so they could evaluate the thing properly first. NSU’s chief executive, Gerd Steiler von Heydekampf, treated the offer as an insult. We don’t take tips, he said.

So Curtiss-Wright found leverage instead. NSU’s American patent filings had been badly handled by an attorney who did not know United States patent law, and Curtiss-Wright pointed out that they could simply build rotary engines in America without a license at all. On October 21, 1958, they signed for $2.1 million — about $24.3 million in 2026 — plus a 5 percent commission — exclusive rights in the United States, plus the right to sell licenses on to other American companies, though NSU’s permission was needed before licensing anyone who built cars or motorcycles. Every American rotary story after that runs through that contract.

Then Hurley went to market — full-page advertisements, film of a small rotary pumping water for fire hoses, expansive predictions — and Curtiss-Wright stock rose about 50 percent. When the dividend was cut soon after, Hurley found himself explaining to the Securities and Exchange Commission that the Wankel was a real engine and not a device for moving his own share price. In Germany it was wilder still: speculators drove NSU stock to 3,200 percent of par value, against the 300 to 400 percent a normal industrial share traded at.

What Curtiss-Wright then built, over about a decade, is the strongest single argument for the rotary’s versatility that anyone ever assembled.

At the small end: an air-cooled lawnmower engine of 4.3 cubic inches making 3.5 horsepower. At the large end: the RC-6 — Curtiss-Wright’s 60-cubic-inch single-rotor benchmark, the RCI-60 family from which the rest of the American program was scaled — blown up 30 times in its basic dimensions, displacing 1,920 cubic inches and producing 872 horsepower at 1,525 rpm. In between they built the 4RC-6, four of those baseline rotors in one engine and the first multi-rotor Wankel ever made, good for 425 horsepower — which led directly to a four-rotor air-cooled aircraft engine. They put a twin-rotor in a 2½-ton Reo Army truck. They built a portable generator around one. They developed marine versions.

And they built the DISC engines — direct injected stratified charge — which used spark ignition with high-pressure injection to run on JP4 jet fuel, JP5, number 2 diesel or high-octane gasoline. Power was lower, but the engine would burn essentially anything you poured into it, which for a military customer is the whole point.

The car, though, is the detail that stops people. Curtiss-Wright dropped a twin-rotor into a 1966 Ford Mustang, replacing the 200-horsepower 289 V8 with a rotary rated at 185 horsepower at 5,000 rpm. They kept the Ford automatic, adapted the Ford alternator and ignition, and fitted a Buick carburetor. Jan Norbye of Popular Science was given an extended loan of it, and Curtiss-Wright would not let him print what car he was driving — Ford had shown no interest in the Wankel, and the rumor would have been awkward for everyone. Norbye, who had driven the supercharged BRM V16 and sat in Ferrari’s test house with four V12s at full song, wrote that the Wankel was a sound he had not heard before: something turbine-like in it, but pulsing like a good six in perfect tune, and giving no indication at 6,000 rpm that it had any intention of stopping.

There was one more near miss. In 1968 Andy Granatelli — race team owner, oil-additive magnate, the man who had just won Indianapolis with Mario Andretti after campaigning turbine cars there for years — approached Curtiss-Wright about buying several four-rotor Wankels to run at the Speedway. Management could not see what a rotary at Indianapolis would be worth to them, and the deal died. It is the largest piece of publicity the American rotary never got.

The big military contract Curtiss-Wright had been waiting for never came. Bentele left for AVCO Lycoming in 1967. In 1984 the company sold its entire rotary business — technology, licenses and all — to John Deere for $14 million (about $45 million in 2026) — only $2 million of it down, the rest in installments — a substantial loss against a decade of development. Deere spent the eighties chasing heavy equipment, aircraft and military work with it, then sold the program on again in 1991, to Rotary Power International, along with 32 years and more than $200 million of accumulated development (about $490 million in 2026).

France had every reason to win this

Citroën should have been the one. This was the company already building cars that rode on pressurized nitrogen spheres at every corner, front-wheel drive across the whole range decades before anyone else bothered, and the DS in 1955. A radical engine was entirely in character.

They went at it structurally rather than casually. In 1964 Citroën and NSU formed a joint company, Comobile, to work out how to put an NSU rotary in a Citroën. In 1967 they formed a second one, Comotor, capitalized at $1 million and later raised to $4.04 million — about $10 million and $40.4 million in 2026 — to build and sell rotary engines for any application at all. In 1969 Comotor bought land in the Saar for an engine plant.

The first product was the M35, and the thinking behind it was genuinely unusual. Citroën decided the only way to learn how the engine behaved in the real world was to sell it to the public and let the public abuse it. So the M35 was an experimental car that private buyers could actually buy, backed by a two-year guarantee with no mileage limit — a guarantee designed to make sure the cars came back for evaluation. It was a small four-seat coupe using body parts from the Ami 8 and the 2CV, with a single-rotor engine hanging out ahead of the front axle and Citroën’s hydropneumatic suspension underneath.

Against the 26-horsepower flat twin in a 2CV, the M35’s 55 horsepower and 50.6 lb-ft were a transformation. It would hold 80 mph in third if you asked it to, the tachometer read to 7,500 rpm, and the engine revved so freely that Citroën fitted a buzzer.

The real one was the GS Birotor: a twin-rotor Comotor engine of 107 horsepower at 6,500 rpm, mounted transversely, driving through a Citroën-built semi-automatic three-speed, in a sophisticated four-door aimed squarely at the same buyers as NSU’s Ro 80. Top speed 110 mph.

It failed completely. The twin-rotor broke apex seals and seized. At $4,500 in 1974 — about $30,500 in 2026 — it was an expensive car, and Comotor’s obsessively precise assembly process — ultrasonic cleaning between machining stages, rotors sorted by weight to within 20 grams, apex seals stocked in eight lengths varying by ten-thousandths of an inch so the assembler could pick a perfect fit — made the engines beautiful and costly, which is very likely why no other manufacturer ever bought one. Citroën recalled most of the Birotors and destroyed them, and went on selling the ordinary piston-engined GS quite happily.

By then the alliance was finished anyway. Relations with NSU had soured through the sixties, a merger was discussed and abandoned, and in 1968 it was Fiat that bought a stake in Citroën from the Michelin family. NSU lost interest entirely.

Germany hedged its bets in writing

Mercedes-Benz did not stumble into the rotary. It made a calculated bet, and the shape of that bet is the single best evidence of how seriously the industry took this engine.

It began as a handshake. Wolf-Dieter Bensinger, who had worked with Wankel during the war and was now responsible for Mercedes passenger-car engines, went to see him in 1960 to discuss replacing the company’s overhead-cam six with a rotary. There was no license and no money at first — NSU supplied test engines, Mercedes agreed to share what it learned, and Bensinger set about building an internal case before asking anyone to spend.

In October 1961 Mercedes paid $750,000 — about $8.4 million in 2026 — for the right to build gasoline Wankels of 50 horsepower and above. Then, in March 1964, they bought a second license: the right to produce diesel Wankels of any size at all. Whether or not anyone at Mercedes had a diesel rotary in mind, the agreement was understood as insurance — protection against technical obsolescence if the rest of the world went rotary and Mercedes had not.

That is the tell. The most conservative engineering company on earth bought an option on a future in which its entire engine philosophy was wrong.

The engineering that followed was serious. Bensinger’s engines ran well for about 400 hours and then lost power, and teardowns always showed the same chatter marks on the housing. His rotor cooling used the rotor’s own motion to swirl oil through its three internal cavities. His combustion research turned up something genuinely surprising: gas velocity in a Wankel chamber is so high that it can outrun the flame front, meaning the apex seals stay relatively cool and can be made from materials with poor heat resistance. And it was Heinz Lamm, running Daimler’s new rotary department, who found the answer to the housing surface — a nickel and silicon-carbide alloy developed by Mahle specifically for Daimler-Benz and sold as Nickasil, which NSU and most other licensees eventually adopted too.

Then came the C111, and it is still one of the great might-have-beens. Conceived by Rudolf Uhlenhaut, who had been central to the Mercedes racing cars of 1937 to 1954 and was frustrated by the company’s withdrawal from the sport, it was a mid-engined plastic-bodied prototype with gull-wing doors, styled by the man responsible for the 300SL. It was first shown at Frankfurt in 1969, badged C101.

Journalists loved it. It became a fixture of the European show circuit. Prospective buyers sent Mercedes blank checks asking to be first in line if it reached production. Mercedes would not say what it intended, denied it was going racing, and eventually walked away — not because the engine had failed, but because nobody knew yet whether a Wankel could be made clean enough to sell in the United States, and the United States was where the money was. Having got the injection systems working and the seals lasting, Mercedes decided to leave the rotary as a laboratory toy for a while.

Britain went after the hardest version of the problem

Rolls-Royce belongs in this story, but not the Rolls-Royce most people picture. The relevant part of the company was the one that built aircraft engines and diesel truck engines, and in the sixties it was full of engineers who wanted to try new things and had the connections to get the British military to pay for the trying.

A technical adviser brought a Wankel into the lab. It reached the chief executive, Sir Denning Pearson, who showed it to a young engineer named Fritz Feller, then working on a multi-fuel engine for tanks and military trucks. Multi-fuel mattered because a military vehicle that will run on anything from jet fuel to gasoline can keep moving when supply breaks down — or use captured fuel. Feller and Pearson saw a rotary diesel: high power for its weight, and ideal for vehicles that had to be flown in.

Getting the license took some doing. The German “Diesel Club” — MAN Krupp, Klöckner-Humboldt-Deutz and Daimler-Benz — had all taken licenses and agreed to share developments, and had no interest in admitting a British firm they were sure would contribute nothing. The genteel British negotiators were unprepared for the German style; at one point a bargaining session in an English country inn dissolved into a shouting match between two of the German negotiators. Rolls got a specialized license in 1965, for engines of 100 to 850 horsepower.

Mercedes and Curtiss-Wright had built spark-ignition engines that would burn diesel fuel. Rolls wanted an actual diesel — compression ignition, with the enormous pressures that implies. The problem is geometric: the K factor needed for diesel-like compression makes the engine enormous. Feller’s answer was to stack two rotary units on top of each other, the larger acting as an air pump for the smaller, which was the engine proper. That let the small unit reach diesel compression without the crippling proportions.

What came out of that arrangement was the 2-R6: a compound diesel of two rotary units, the larger breathing for the smaller, and the engine the whole Rolls-Royce program was building toward.

They redesigned the apex seals into an L shape with passages that let gas pressure push them outward, which is the sort of fix that only shows up when you run an engine at pressures nobody intended. And here is the irony that ties the whole story together: the ceramic silicon-carbide seal material Rolls developed for its diesel turned out to be a perfect match for the Nickasil housing surface Mercedes had settled on. Mercedes, in the business of expensive cars, cheerfully paid about $20 worth per twin-rotor engine, roughly $135 in 2026 money. NSU — the company that owned the patent everyone was licensing — looked at the price and refused to buy it.

The 2-R6 never got its chance. Rolls-Royce went bankrupt, Pearson resigned, the British government nationalized the company, and the rotary program was officially stopped in 1974. There is also a plausible outside cause: after the Yom Kippur War of 1973 demonstrated what portable anti-tank missiles could do, the British military rethought the airborne tank, and the tanks the Rolls rotary was meant to power were dropped from procurement.

And now the part almost nobody knows

Motorcycle manufacturers had been watching the whole time, and several of them bought licenses.

The first production rotary motorcycle was the Hercules, sold as a DKW for export, using an air-cooled 20-horsepower engine from Fichtel & Sachs mounted with its shaft running fore and aft, six speeds and shaft drive. Early bikes did 82 mph; a later 32-horsepower version reached 93. Only about 1,800 were sold, which by any commercial measure is a failure. But it worked, and when it ended, all of the Wankel manufacturing equipment went to the Norton Motorcycle Company of England in 1977.

Norton had already started. A development engineer named David Garside had put a Fichtel & Sachs engine into a BSA 250 Starfire chassis and liked how it went, but hated the cooling. Sachs used the incoming intake charge to cool the rotor, which is clever, but the power lost to heating the charge was more than Garside would accept — so he rerouted the intake so the air passed through the rotor, picked up heat, then went out into a plenum to cool back down before entering the engine. Bert Hopwood, a retired Triumph engine designer, joined in, and together they worked up a twin-rotor with several times the cooling fin area.

The production Norton was not beautiful. Cook Neilson in Cycle called it an alien lump of aluminum with arcane internals and said its appearance would appeal mainly to air-conditioner salesmen. But it grew on people, it went extremely well, and in a country raised on vertical twins the absence of vibration came as a relief.

That is worth sitting with. The rotary engine — the commercial disaster, the thing that defeated Citroën and Mercedes and Rolls-Royce — is what kept Norton breathing after the rest of the British motorcycle industry had gone.

Suzuki started work on a motorcycle rotary in 1970 and launched the RE-5 in 1975, at exactly the moment the Japanese industry was done copying British designs and reaching for anything genuinely new. They hired Giorgetto Giugiaro to style it, and the folded-paper precision that would give the world the Lotus Esprit a year later did not survive the translation to two wheels. The instruments sat in a cylinder shaped like a Thermos flask with a flip-up lid, and the bodywork was so relentlessly circular in tribute to the engine that collectors call the early bikes the Wurlitzer model.

The joke is that the rotary was chosen for motorcycles because of its power-to-weight ratio, and the RE-5 was neither powerful nor light: 48 horsepower and 260 kg (573 lbs). The engine was not the problem. The equipment needed to run it was — a radiator, plumbing and coolant, plus a double-walled exhaust system, because exhaust gas came out 300 to 400 degrees hotter than a conventional engine’s. The engine itself was real progress: all aluminum, peripheral port induction, chrome-plated trochoid housing, a four-piece self-adjusting apex seal arrangement, and a carburetor so complicated it took five cables to operate.

Then there is the Van Veen, which is the one to bring up if you want to end an argument. A Dutch outfit put a Comotor twin-rotor — the Citroën-NSU engine, the same family that powered the GS Birotor — into a Moto Guzzi frame. It made 110 horsepower, weighed 281 kg (620 lbs), did more than 130 mph, and cost £5,000 (about $63,015 in 2026 money). They built them from 1976 to 1978.

Yamaha got as far as a prototype, the RZ201, built under a Wankel license held by their diesel subsidiary Yanmar. A 660cc liquid-cooled engine, shown at the 1973 Tokyo Motor Show. It never reached production.

Lawnmowers, snowmobiles and a boat that kept losing

Wherever fuel economy did not matter much, the rotary kept turning up.

Outboard Marine Corporation built a racing outboard that ought to have won everything. A witness to several regattas described the pattern: the rotaries would lead the field for about two-thirds of the race, then the seals would go off and the boats would slow, and slow, and finish somewhere behind everybody, if they finished. That is the rotary’s whole biography in one afternoon on the water.

Arctic Cat built snowmobiles with Fichtel & Sachs engines. OMC built engines for Johnson and Evinrude snowmobiles. Through 1974 and 1975 you could find rotaries in lawnmowers, water pumps and stationary generators. A few people are still keeping the last rotary snowmobiles running by cannibalizing the others.

And then a cork company did it

Toyo Kogyo means, roughly, Oriental Cork, and that is exactly what the company founded in Hiroshima in 1920 made at first. Then bicycles. Then motorized bicycles. Then, in the thirties, three-wheeled delivery trucks that had more in common with a motorcycle than a car and looked rather like a Harley-Davidson Servi-Car. They were called Mazdas, after the founding family’s name, Matsuda. By 1940 the company was selling up to 1,500 of them a year.

The factory sat outside Hiroshima, separated from the city by a small range of mountains. In 1945 that geography saved it. Some glass broke and some roofs were damaged; the buildings and the people inside them survived. Matsuda later observed that the employees who died that day were the ones who had not come to work.

Kenichi Yamamoto graduated in 1944, loved aircraft, and spent the end of the war designing landing gear for fighters — including, finally, a kamikaze aircraft that was never meant to land at all. He came out of it unemployed in a region where the industry had been flattened, and went to work at the only surviving factory, building gearboxes for three-wheeled trucks on an assembly line. He found it boring, so he asked permission to study the drawings on his supervisor’s desk and check them against the parts he was making. Somebody noticed, and he was moved to engine design.

By 1959 Tsuneji Matsuda, the founder’s son, was running the company and looking for something that would stop Toyota or Nissan swallowing it. The rotary looked like exactly that — but getting a license meant standing behind roughly 100 other companies, and Toyo Kogyo only got through the door because Matsuda had befriended the West German ambassador to Japan. In October 1960 he took five technical staff to Germany, where NSU’s engineers balanced a coin on its edge on a running engine to demonstrate how little it vibrated. The agreement was signed on October 12.

What they got for their money was a KKM400 prototype, some blueprints and a great deal of hope. Their own copy shook, smoked, drank oil and failed after 200 hours on the stand, with chatter marks scored across the housing. Yamamoto’s view — and this is the sentence the entire rotary story turns on — was that the engine was not inherently flawed; the metallurgy and machining of the day simply had not caught up with it.

By the time Yamamoto’s team went from their first prototype in 1961 to their first production car, they had piled up nearly 5,000 junk rotary engines in the back lot behind the workshop.

They built a test laboratory with 30 cells running endurance tests around the clock, computer-logged. They chased chatter marks by drilling holes through the base of the seals to change their resonant frequency, which worked and was too delicate to produce. They invented a coating process that sprayed steel into the casting die before pouring aluminum, because chrome would not stick to aluminum but would stick to steel. They redesigned the phasing gear teeth after engines started tearing themselves apart on the stand — one engineer said they behaved as though they were having a fit. They found that a hard-driven cold engine hit 450°F inside the housing within a minute of starting and cracked around the spark plug openings after 4,000 cycles, and they redesigned the housing until it stopped.

Toyo Kogyo spent more than $50 million on the rotary through the sixties, better than $530 million in 2026 money. By 1970 the company was running on the ragged edge, kept solvent only just by its ordinary piston-engined products, and anything short of total success with the rotary cars would have meant being absorbed by Toyota or Nissan. They went to the United States because it was the only market big enough to save them.

And then they went racing

Everything above is a list of companies that stopped, and every one of them stopped on the same point: the engine would not last. Mazda spent the next two decades arguing the opposite in public, on a clock, in front of witnesses.

The Cosmo Sport made the case first. In 1968 Mazda entered two of them in the Marathon de la Route, 84 hours around the Nürburgring, and one came home fourth overall with nothing but Porsches and Lancias ahead of it. For an engine that had been scoring chatter marks into its own housings a few years earlier, three and a half days of the Nordschleife was not a result. It was a rebuttal.

Then the RX-7 turned the rebuttal into a career, and the numbers are faintly absurd. It won more IMSA races than any other model of car from any manufacturer — 100 of them, the hundredth at San Antonio on 2 September 1990, 12 years after the first. It won its class at the 24 Hours of Daytona 10 years in a row from 1982, and gave Mazda 8 consecutive GTU manufacturers’ championships between 1980 and 1987, which took the most-championships record off Porsche. In SCCA club racing, cars running Mazda engines have won more than 100 amateur titles. Hege calls the RX-7 the most successful race car ever built, and it is hard to argue him down. The road cars kept step: the second generation arrived in October 1985 with the larger 13B, an intercooler and a twin-scroll turbo, the third in 1993 with sequential turbochargers that cut the time to full boost by 43 percent, and by 1990 the three-rotor was in production, turbocharged, in the Japan-only Eunos Cosmo.

Le Mans took longer. A Belgian privateer put a 10A in a Chevron B16 in 1970 and lasted 4 hours before a cooling hose let go. A Japanese Sigma with a 12A stopped at 11 hours in 1973; the next year’s car was still running at the end but had not covered enough laps to be classified. It was 1980 before a Mazda rotary actually finished the thing, an American-entered RX-7, 21st overall. Then the climb: class wins with the 717C and the Lolas, the GTP category in 1987, and in 1989 three cars filling the top three places in GTP after two of them had been crashed in qualifying and repaired in time to start.

Which brings us to 1991, and the best hour the rotary ever had. Race officials had told the teams that rotaries would be disqualified the following year, so Mazda understood this to be its last chance at an outright win and built for exactly that: the fuel consumption that had killed the 787s sorted out in the R26B, the carbon fiber body stiffened. The car weighed 831 kg (1,831 lbs) and the four-rotor made 700 horsepower — and that engine was nothing more exotic than the RX-7’s 13B taken to four rotors, 654cc a chamber, the same arithmetic that gives 1,308cc in a road car. Number 55 wore orange and green for a Japanese clothing company called Renown, and it did not sound like anything else that has ever been to Le Mans.

It sat third at half distance. It ran with the front group for another 9 hours. Then Mercedes came into the pits, Mazda went by, and it never gave the lead back. At the finish the two 787Bs were first and sixth overall, and the cars behind them carried the badges of Jaguar, Mercedes-Benz and Porsche — every one of them running pistons, every one of them backed by money Toyo Kogyo had never had.

The engine that beat them was the one every last one of them had already given up on.

Curtiss-Wright had sold its rotary to a tractor company. Citroën had crushed its cars. Rolls-Royce had gone bankrupt trying. Mercedes had put the C111 in a museum and walked away. And the company that finished the job — that took a German patent nobody else could keep alive and made it win the hardest race there is — was the one that started out in Hiroshima making cork, then bicycles, then three-wheeled trucks. It is still the only overall victory at Le Mans by a car without pistons. It was also the first by a Japanese manufacturer, and it stayed the only one for 27 years, until Toyota managed it in 2018.

The legend says the rotary was banned afterward for making a mockery of the piston engines. It is a good story and it is not true. The 3.5-liter formula that would have made the 787B illegal was settled before the race, for reasons that had nothing to do with Mazda — and then the disqualification never actually happened at all. Rotaries stayed legal. Mazda kept entering. The factory simply never spent like that again. Nobody slammed the door on the rotary at Le Mans. It won the biggest race in the world, and then the room quietly emptied around it.

Still turning

By the turn of the century Mazda was the only carmaker on earth that still owned the right to build and sell Wankels, and it had no intention of giving that up.

The hydrogen work had been running since 1991, when Mazda showed a hydrogen rotary at the Tokyo Motor Show. The clever part was not the engine but the tank: a metal hydride that held the gas inside the crystal lattice of the metal itself and released it as the engine’s own coolant warmed it. For real-world testing they fitted hydrogen rotaries to Miatas. Mazda’s argument for the pairing is a sound one — the rotary’s combustion cycle is slower, so fuel and air have longer to mix, and there is no exhaust valve sitting there glowing red and lighting the charge early.

Then, at the 1999 Tokyo show, the RX-Evolv: a four-door with rear doors hinged at the back like a 1961–1967 Lincoln Continental, which Mazda cheerfully called a family sports car. Under it was the Renesis, and the Renesis did something the company had tried in the early days and abandoned. It moved the exhaust ports off the periphery of the housing and onto the side housings, below the intakes — the layout the engineers had wanted all along and could not seal in the sixties. With the ports moved and the rotors lightened, the engine would turn to 10,000 rpm and made its peak at 9,000 instead of 6,500. Hege’s verdict on that: Mazda’s rotary was at last becoming the race horse Wankel had envisioned, rather than the cart horse he had labeled it.

The RX-Evolv became the RX-8 in 2003, Renesis and all, and Mazda built roughly 200,000 of them before emissions rules finished the car in 2012. The rotary then left the road entirely for 11 years. When it came back, in 2023, it came back in the MX-30 R-EV as an 830cc single rotor that does not drive the wheels at all — it sits under the hood generating electricity for a battery, which is to say the engine had gone all the way around to doing the job Wankel’s very first rotary did: feeding a motor rather than being one. Mazda ended production of that car in January of this year.

The engineers are still at it, though. There is a dedicated rotary development group at Mazda again, 36 people whose entire job is this engine, and two cars to point at: the Iconic SP concept of 2023, and the Vision X-Coupe shown at the Japan Mobility Show last October — a turbocharged twin-rotor plug-in hybrid making 503 horsepower, this one allowed to drive the wheels, burning a carbon-neutral fuel brewed from algae. Whether either ever reaches a showroom is anyone’s guess; Mazda has been making that promise, on and off, for 25 years.

The point

The usual telling makes Mazda’s success look like stubbornness rewarded, and it was that. But it looks different when you know what it was measured against.

Citroën had more genuine innovation in its ordinary showroom cars than almost any company alive, and threw two cars and two joint ventures at the rotary before recalling the results and crushing them. Mercedes had the money, the engineering depth and a prototype quick enough that people mailed in blank checks — and filed the whole thing away. Rolls-Royce had the aircraft expertise and the British military behind it, and got as far as bankruptcy. Curtiss-Wright built rotaries from lawnmower size to 872 horsepower, put one in a Mustang, and sold the lot to a tractor company at a loss.

Everybody tried it. The engine went into cars, trucks, boats, snowmobiles, generators, lawnmowers, aircraft designs, an army truck and a Ford Mustang. And the ones that lasted longest in production — the ones almost nobody remembers — were the motorcycles: a German export bike that sold 1,800 copies, a British police bike, a Suzuki with a Thermos flask on the handlebars, and a Dutch superbike carrying a French-German engine in an Italian frame.

All of that, from a supercharger bolted to a moped.