The Chevrolet Corvair's air-cooled flat-six, 145 cubic inches, no radiator, and why Porsche's own engineers could not copy the part that cooled it.
Most men who owned one remember the fan belt that came off above four thousand rpm, and the book that ended the car. The story everyone repeats is that Chevrolet did air cooling roughly and Porsche did it properly.
This documentary starts in 1957 at Warren, Michigan, where Chevrolet put its first flat-six into a borrowed Porsche 356, and follows the car through Stuttgart, where Ferry Porsche drove one of the first Corvairs built and Porsche tested its own flat-six under Corvair wagons. On the way: Ed Cole's Continental tank engine, the Holden paperwork, Al Kolbe's 1958 patent, the Massena foundry, Tonawanda, Willow Run, and the last car off the line on May 14, 1969.
CHAPTERS 0:00 What Did Chevrolet Do With A Porsche? 2:58 Why Couldn't Porsche Make The Number? 9:18 How Did Porsche Cool Its Six? 12:25 What Sits On Top Of A Corvair? 15:24 Where Did The Flat Fan Come From? 18:31 What Did The Fan Really Cost? 21:41 How Did Cole Get It Approved? 27:52 Why Did The Belt Come Off? 30:58 What Did Stuttgart See? 39:50 What Could Porsche Not Copy?
This analysis explores the divergent engineering philosophies of Chevrolet and Porsche in developing air-cooled flat-six engines. While Porsche prioritized precision machining and high-cost castings for luxury performance, Chevrolet leveraged industrial stamping and mass-production tooling to make air-cooling viable for the family sedan market.
In the early summer of 1957, Chevrolet Engineering bought a Porsche 356 1600 Coupe, dropped the
engine out of it, and put that engine on a dynamometer to read temperatures and clearances
under instruments.
Here is the part that does not fit.
The answer the record finally gives is not inside that engine at all, and it is not a
part Porsche ever had to build.
It is not the engine.
It is what carries the air to it.
Late in 1957, at Chevrolet's new engineering center at Warren, Michigan, that same 356
was still in the building.
Its own flat four was gone.
In its place, bolted up to a Chevrolet manual transmission, sat the first running Corvair
flat six, because the engine was ready for road work before any Corvair prototype was.
Edward N. Cole, who ran Chevrolet Engineering, got into it and drove.
He drove it at the Technical Center and at Milford Proving Grounds in late 1957, and
at Pikes Peak, Colorado, in early 1958.
Cole later told the historian L. Scott Bailey what that felt like.
The first concrete evidence the thing came up to their hopes, a modified Porsche carrying
the new Chevrolet engine, and in his words, she ran beautifully.
The first flat six-powered Porsche in history was built by Chevrolet, in Michigan, and driven
by an American engineer who had logged about 300 hours in a Continental-powered Bonanza.
That is where this begins, and it is not where the familiar story goes.
This is not a ranking, and it is not a nostalgia reel.
It is also not the Nader story.
Ralph Nader's asterisk, unsafe at any speed asterisk, appeared in the autumn of 1965,
and it changed everything commercially.
And that single clause is all it gets here, because it answers a different question from
the one on the table.
The question here is narrower and harder.
Chevrolet sold air-cooled six-cylinder cars in America by the hundred thousand, finishing
at 1,786,243 built in the United States when the last one came off the line on the 14th
of May, 1969.
Porsche, the company with more air-cooling experience than anyone on Earth outside Wolfsburg,
did not have a flat six on sale anywhere until 1965.
And when its engineers looked at what Chevrolet had done, the specific thing that made it
possible was not something they could copy.
So what was it?
It is not a secret, it is not a trick, and it is not really an engine.
It belongs to a category most people never think about when they think about engines
at all.
Production tooling.
Specifically, a kind of tooling Detroit had in enormous quantity for a reason that had
nothing to do with cooling anything.
Attach one number to it and hold it.
1,786,243.
The rest of this is the case.
It starts in Stuttgart, with a man under orders he could not meet.
Klaus von Rieker had a target, and he could not reach it.
Von Rieker was Porsche's technical director at the turn of the 1960s, and the job in front
of him was the successor to the 356, a car that would eventually be called the 901, then
the 911.
Ferry Porsche's requirements were not negotiable, and they were not numerous.
The car would keep its engine behind the rear axle.
It would stay air-cooled.
And the engine would make 130 horsepower by the German DIN standard, which works out to
roughly 148 horsepower by the gross American measure used in Detroit advertising at the
time, because DIN counts the engine with its fan, generator, and exhaust attached,
and the American gross figure did not.
That last distinction matters more than it sounds.
When a German engineer quoted output, the cooling fan was already eating into the number
he quoted.
When an American engineer quoted output, it often was not.
Two engines with the same figure on paper were not the same engine, and everyone in
both buildings knew it.
Von Ricker's answer was the Type 745, a pushrod air-cooled flat six.
It did not make the number.
The accounts of exactly how short it fell vary, but the outcome is not in dispute.
The 745 did not satisfy Ferry Porsche, and in 1962 Von Ricker left Porsche for BMW.
Hans Tamala took over engineering on the 901 project, and in January of 1963, a young engineer
named Hans Mesger began the design that became the flat six the world knows.
None of this was incompetence.
Von Ricker was working inside a constraint most people never name when they tell this
story.
Porsche built cars in the hundreds and low thousands.
It had a foundry relationship, a machining capability, and a workforce sized for that.
Every gram of that engine was going to be made in small numbers by skilled hands, and
every solution had to be one that stayed economic at small numbers.
There is a phrase that circulates about the standing order to keep air cooling at Zuffenhausen,
Shoemaker, stick to your last.
Whether Ferry Porsche said it in those exact words is not something the record settles
cleanly.
What the record does settle is the behavior.
Porsche kept air cooling in production cars for another 35 years after the 901 through
1998, and did not abandon it until emissions and noise law made it impossible.
So here is the shape of the trap Von Ricker was in.
He had the correct layout, he had the correct cooling philosophy, he had the best air cooling
engineers alive, and he was being asked to extract a specific horsepower figure from
a configuration where, unlike a water-cooled engine, a meaningful share of the power he
made had to be spent immediately on making the cooling work at all.
That share is not a small number, and nobody at Chevrolet was advertising it either.
To understand why it exists, you have to look at what air physically cannot do.
Put your hand flat on a cold metal gatepost on a winter morning and it hurts within seconds.
Stand next to the same gatepost in air of exactly the same temperature and nothing happens.
The air is just as cold.
It simply cannot move heat out of your hand at anything like the rate the metal can.
That is the entire problem, and it is the reason the Corvair and the 911 exist in the
shapes they do.
Heat leaves hot metal into a moving fluid at a rate governed by how much heat that fluid
can carry and how fast it flows past.
Water carries heat away from a hot surface something on the order of a hundred and seventy-five
times better than air does, for the same surface, the same temperature difference, and the same
flow speed.
That is not a rounding difference.
That is a different problem.
An air-cooled engine has to claw back that gap, and it has exactly three ways to do it.
It can present far more surface area to the air, which is what fins are, a Corvair cylinder,
and head give the air roughly ten times the surface a bare water jacket would need.
It can run a much larger temperature difference, letting head metal sit far hotter than a water-cooled
head ever would.
So a Corvair's cylinder head might be operating three or four hundred degrees above the incoming
air where a water-cooled head lives perhaps a hundred and fifty above its coolant.
And it can move air violently, not the gentle trickle a radiator needs, but a hard pressurized
blast forced between fins that are only a few millimeters apart.
Chevrolet built a warning light and a buzzer into the Spyder and Corsa dashboards that
Sit with that.
The factory expected an owner to be watching a gauge measuring the temperature of the metal
around his spark plugs, and it fitted a buzzer in case he was not.
No water-cooled Chevrolet of 1962 had anything remotely like it, because no water-cooled
Chevrolet needed one.
All three of those compensations cost something, and the third one costs the most.
Forcing air through tight fins against resistance is a compressor's job, and compressors take
power off the crankshaft to do it.
But the hardest part is not the total heat, it is where the heat is.
On a flat six, the two center cylinders sit in the middle of the block, boxed in on both
sides by their neighbors, with the least clear path to fresh air of any cylinder in the engine.
On a water-cooled six, that is unremarkable.
The pump shoves coolant everywhere, and it evens itself out.
On an air-cooled six, the center cylinders are the ones that will let go first, and the
record is explicit that Chevrolet's designers were having, in the plain phrasing of the
account, special problems cooling the center cylinders.
That was the problem the borrowed Porsche engine on the Warren dynamometer could not
help them solve.
Porsche's engine had four cylinders.
It did not have a middle.
9.8 inches across, 11 blades, spinning in a fiberglass duct sitting on top of the engine
at 1.3 times crankshaft speed.
That is the cooling system Hans Metzger's team arrived at for the 901, and it is one
of the most elegant pieces of packaging in the history of the motor car.
The fan stands upright, axially, drawing air in and driving it down over six separately
thin cylinders and six separate thin cylinder heads.
Because each head is its own casting, air reaches all six equally, there is no center
cylinder hiding behind its neighbors the way there is on a shared casting, and there is
no coolant jacket to distribute anything, so the geometry has to do the job on its own.
The alternator lives inside the fan hub, driven by the same belt, which means the accessory
drive and the cooling drive are one object rather than two.
The cylinders came from MAHL, beryl construction, which means a cast iron liner with an aluminum-finned
barrel cast around it, so the bore surface is hard iron and the fins that shed the heat
are aluminum, which conducts roughly three times better than iron does.
That is an expensive way to make a cylinder.
It is also, for the job, close to the best way anyone had found.
Every one of those decisions is correct.
Every one of them is also a decision available to a company building cars in the low tens
of thousands a year with a German supplier base and a customer willing to pay sports
car money.
Now look at what it was answering.
Mesger was solving for 130 DIN horsepower out of two liters in a car that would be sold
to people who cared about the number.
He was solving for durability at sustained high speed on unrestricted roads.
He was solving for a 10 or 15-year production life at volumes Porsche could machine.
What he was not solving for, because nobody asked him to, was how to put that same cooling
architecture into a four-door family sedan that had to leave a Chevrolet dealership at
a price close to a full-size Biscayne in quantities of a quarter of a million a year.
The 901 engine's answer to air cooling is a precision-machined, supplier-intensive,
individually cast answer.
It works beautifully.
It is also, by construction, uncopyable at Chevrolet's volumes and Chevrolet's price.
This is the point in the case where it becomes clear the two companies were not in a race.
They were answering different questions with the same physics, and the answer that survives
in the popular story that Porsche did air cooling properly and Chevrolet did it roughly
is not what the hardware says.
The hardware says Porsche bought its way past the problem with precision, and Chevrolet
bought its way past the problem with something else entirely.
Something you can hold in two hands that cost almost nothing, and that everyone who has
ever pulled a Corvair engine has cursed at.
It is a shallow steel pan, roughly the footprint of the engine it sits on, with a large round
opening in the middle, a scalloped notch cut on each side, a square hole at one corner,
and weld nuts dotted around its flange.
It weighs very little.
Owners call it the turkey roaster, because that is precisely what it looks like, the
lid off a large roasting tin.
Nothing about it looks like the answer to anything.
Here is what it actually does.
The fan sits in that central opening and throws air outward, and the pan turns the whole top
of the engine into a sealed plenum, a pressurized box with only one way out.
The air cannot go up, because the steel is in the way.
It cannot escape sideways, because the pan's flanges meet the lower sheet metal around
the cylinder banks.
The only rot left is down, through the fins on the cylinders and heads, and out through
the ducts under the engine.
Without that pan, the fan is simply stirring the engine bay.
With it, the fan is a pump, and the engine is the restriction.
The scalloped notches on each side are there because Chevrolet put a carburettor on each
bank, rather than one in the middle, a decision Bob Benzinger, who designed the engine's crankcase
ventilation, explained as keeping the carburettors as close as possible to the inlet ports for
response and volumetric efficiency.
The square opening is for the oil cooler, which on a Corvair sits up in the airstream
inside that same plenum, because the oil is doing cooling work the air cannot do alone.
Spark plug guide tubes pass through it.
Heater ducting bolts to it, because on an air-cooled car the cabin heat comes off the
engine's own cooling air.
Now the part that matters.
That shroud is a steel stamping.
It was pressed out of sheet on a die, weld nuts added, painted, boxed, and shipped from
a stamping plant to the engine line at Tonawanda.
It was not cast, not fabricated, not laid up.
The part numbers run in the 378-0000 and 39-0000 series through the decade, and the enthusiast
archives that have documented the variants have found them changing repeatedly.
New notches, added ducts, a belt guard, while the basic pressing carried on.
This is the first place the shape of the answer shows through, so read the next sentence twice.
The most sophisticated element in Chevrolet's air-cooling system is not machined, not cast,
and not precision-toleranced.
It is a piece of pressed sheet steel, and the reason Chevrolet could build one for every
one of 1,786,243 cars is a reason that has nothing to do with engines.
But the idea of blowing air downward through a flat engine from a flat fan sitting on top
of it did not start at Chevrolet, and it did not start in a car.
It started in a tank.
14.7 liters, 895 cubic inches, six cylinders lying flat, a four-main bearing crankshaft,
a single overhead camshaft per bank, two valves per cylinder closed by triple springs, a centrifugal
supercharger fed by twin-throat downdrat carburetors on each side, twin Bendix Scintilla Magnetos
firing dual ignition, 500 horsepower gross at 2,800 revolutions per minute, 380 as installed,
torque of 975 pound-feet gross at 2,250 revolutions per minute, falling only to 825 installed
at 2,000.
That is Continental's AOS-895-3, built in Muskegon, Michigan, and it powered a light
tank.
In 1950, Ed Cole, then a senior engineer at Cadillac, was handed the job of developing
a facility to build that tank.
In his own account, given years later, Cadillac could not argue with Army Ordnance about the
engine supplier because Continental had been developing this unit expressly for tank use.
So Cadillac's engineers spent a program's worth of time inside an air-cooled flat-six
they had not chosen.
In Cole's words, all their engineers on the project became very familiar with it.
He also noted the layout's appeal in the plainest possible terms, straightforward and simple.
Look at how Continental cooled it, not with an upright fan at one end like a Volkswagen,
with a fine-bladed flat fan mounted above the cylinders, driving air down through the
cylinder fins and out to the sides through a pair of oil coolers.
Down, through, out the sides.
That is the Corvair's cooling architecture, described a decade early, in a fighting vehicle,
driven by gears and shafts rather than a belt.
Cole's other exposure was in the air.
He logged around 300 hours in a Continental-powered Beechcraft Bonanza around 1950, much of it
on instruments in unpleasant conditions, and he was explicit with L. Scott Bailey about
what that taught him.
The reliability of that engine, which he rated highly then and still rated highly decades later.
Flat-sixes at the time were fundamentally aircraft engines, Franklin, Continental, Lycoming,
and almost nobody built them for cars.
Now the counterweight, because it belongs in any honest version of this.
Chevrolet had tried air-cooling once before and been humiliated.
The 1923 copper-cooled Chevrolet, with copper fins brazed to its iron cylinders, was a
technical and commercial failure so complete that the company recalled and destroyed most
of the few hundred built.
That memory was still inside the corporation when Cole proposed doing it again.
So when the Corvair's cooling fan went on flat, above the engine's center, rather
than upright at one end as Porsche and Volkswagen had done, that was not Chevrolet copying Germany badly.
The fan type came from the German Precedent, a forward-curved centrifugal blower.
The mounting came from the tank.
And the reason for the mounting was the car's low rear deck.
A flat fan on top of a flat engine needs something above it to make the air go where it should.
It also needs power.
Far more than anyone outside Chevrolet's own test cells knew.
Hold one number in your head before anything else in this chapter.
26.
Chevrolet's own engineering work on the Corvair's cooling fan put the power absorbed by that fan
at approximately 26 horsepower at 6,000 crankshaft rpm.
And the engineers expected it to climb steeply beyond that, because the power a fan consumes
rises roughly with the cube of its speed.
Double the fan speed and you do not double the loss.
You multiply it by about 8.
That is why the figure gets frightening past the top of the rev range and why it is almost
invisible at idle.
Set that against what the engine was rated at.
A 1960 Corvair's base engine was quoted at 80 gross horsepower.
The hot 1965 Corsa turbo was quoted by Chevrolet at 180 gross.
A figure some contemporary testers openly suspected was generous.
The four carburettor 140 engine was quoted at 140.
So at the top of the rev range, the cooling system on a Corvair could be eating an amount
of power comparable to the entire output of a Volkswagen Beetle of the same decade.
Not wasted, exactly, bought.
That power is the price of not carrying a radiator, a water pump, a fan clutch, hoses,
a thermostat housing, and something like 20 pounds of coolant.
This is the hidden cost and it is the one nobody puts in an advertisement.
A water-cooled engine's pump takes perhaps one or two horsepower.
An air-cooled engine's blower, working against the resistance of tight fins in a sealed box,
takes an order of magnitude more.
And it takes it worst exactly when you are asking the engine for everything.
Now think about who can afford to pay that.
If you are Porsche, selling on quoted DIN output to buyers who read magazines,
that loss comes straight off the number you are judged by.
And Measure's team had to fight for every unit of it.
If you are Chevrolet, selling an 80-horsepower family sedan to a man who wants to get to work,
the fan loss is real but commercially survivable.
It shows up as a slightly slower car, not as a lost sale.
Chevrolet could pay for cooling and horsepower because Chevrolet was not selling horsepower.
That is the trade in one sentence, and it will matter at the end.
One more thing to carry forward, because it will come back.
On the 29th of November, 1963, a change was directed into the top shroud, a belt guard.
A small piece of pressed metal, added to a part already in production for four model years,
to deal with the single most notorious behavior of this engine.
Hold that date, 29-11-63, because the fan that took 26 horsepower
was driven by a belt that had to turn a right angle twice to get there.
And every engineer who saw it said the same thing,
that belt is going to come off.
They were not wrong.
But first, somebody had to get the car approved at all.
And the way Ed Cole did that was by pretending it was Australian.
The paperwork said Australia.
Through 1956 and 1957, inside Chevrolet Engineering,
a rear-engine compact car was being developed under the name Holden,
GM's Australian subsidiary.
And it was carrying Holden part numbers.
Chevrolet did design work for Holden as a matter of routine,
so requisitions, purchase orders, and tooling inquiries
bearing Holden identification raised no eyebrows anywhere in the system.
Internally, the car went by Holden 25.
Ed Cole, running Chevrolet Engineering,
was developing a car that senior GM management had not approved,
on stationary belonging to a division on the other side of the world.
Understand what that actually required.
Not secrecy in the melodramatic sense,
this was a large engineering organization with clerks, buyers, and model makers,
but a plausible administrative cover that let real money move
and real suppliers be contacted before any committee said yes.
The styling had to be hidden too,
rather than run it through Chevrolet's production studio,
where its existence would have been common knowledge across styling staff.
The work went to Ned F. Nichols, who ran an experimental group,
working under William the 50th.
Mitchell, Harley Earl, was still the styling vice president then.
He left that office in December 1958,
and the Corvair's shape carries his trademark lowness,
even though the drawings were Nichols and Mitchell's.
In September 1957, with Nichols' team having completed the first full-size clay models,
Cole finally showed the whole thing to GM president Harlow Curtis
and explained the project end-to-end.
Curtis was extremely sceptical.
Cole had answers prepared for every question, including the unglamorous ones,
where the production facilities would come from,
where the raw materials would come from.
Roughly three months later,
Cole made the formal case to the Corporate Engineering Policy Committee.
The timing was luck of a brutal kind.
The 1958 Chevrolets had landed to poor response.
The American economy was sliding into recession.
Volkswagen was building a genuine dealer and service network in the United States,
and GM was heading for one of its worst years in memory.
Cole argued that Ford and Chrysler would move on compacts, whether Chevrolet did or not.
He did not know it for certain at that moment, but he was exactly right.
Ford approved the Falcon that spring, and Chrysler approved the Valiant.
He got his approval, rear engine and all.
Meanwhile, the engine itself was being pinned down in detail by Adelbert Kolbe,
universally called Al, who applied for a patent on its layout on the 31st of December, 1958.
Kolbe's engine carried Chevrolet's own stamped steel rocker arms,
first used on the small-block V8 in 1955,
and hydraulic zero-lash tappets to cope with the differential expansion you inevitably get
when iron cylinders and aluminium heads are heated to the temperatures air cooling demands.
The camshaft drive sat at the flywheel end, unlike Porsches and Volkswagens.
An approved car and a patented engine are still only drawings.
Somebody now had to cast 92 pounds of aluminium per engine,
hundreds of thousands of times a year, in a country that had never done it at that scale.
Chevrolet did not know how to do this.
That is not a slight, it is the documented position.
The division had very little practical experience with aluminium engines at the end of the 1950s,
and the Corvair was going to need aluminium in quantities no American car company had handled,
around 92 pounds of it in every single engine,
in a split crankcase and six-cylinder heads that had to hold gas-tight joints
while running hundreds of degrees hotter than any cast-iron Chevrolet block ever would.
So GM went and got the knowledge, and the record on where it came from is specific.
The moulds and foundry procedures were developed with the help of the German firm Carl Schmitt
of Neckarsolm, Carl Schmitt Metal Schmelzwerk in the fuller form,
and the low-pressure casting machines used for the aluminium parts were built
and installed by Carl Schmitt GmbH of Neckarsolm.
Stop on that for a second, because it cuts directly against the folklore.
The story that circulates is that Chevrolet copied Porsche.
What actually happened is that Chevrolet bought German process knowledge,
not German engine design, from a piston and casting specialist 150 miles up the road
from Zuffenhausen, and used it to do something Porsche had no reason to attempt.
Low-pressure casting is worth one plain sentence.
Instead of pouring metal into a mould under gravity, or ramming it in at high pressure,
the furnace is sealed and gently pressurized,
so molten aluminium rises up a tube into the mould from below, filling it slowly and calmly.
You get fewer trapped bubbles and a denser casting than gravity pouring.
At production rates, gravity pouring cannot match,
which is precisely what you need for a part that will be sitting at 400 degrees
with combustion pressure on one side of it.
Then there was the metal itself.
GM built a new castings plant at Messina, in far northern New York, on the St. Lawrence.
And the reason it went there, rather than Michigan,
is the reason anything aluminium goes anywhere – electricity.
Smelting aluminium is essentially a way of turning electric power into metal,
and Messina sat next to cheap St. Lawrence hydroelectric power.
Reynolds Metals operated a reduction plant in that same location.
The castings came out of Messina.
The engines were assembled at Chevrolet's Tonawanda plant near Buffalo.
The unitized bodies were built at Willow Run,
the enormous former bomber factory GM had bought from Kaiser-Fraser in 1953.
Now put the map together, because it is the whole argument in geography.
A German casting process in an American plant cited for hydroelectric power,
feeding an engine line in Buffalo,
feeding a body plant that was built to make heavy bombers.
Willow Run is the tell.
The plant that stamped the Corvair's body panels was a plant designed around enormous presses,
and the first thing that infrastructure had to fix was a belt that would not stay on.
A Corvair owner in the spring of 1960,
pushing past 4,000 rpm on an on-ramp, could hear his fan belt leave.
It is the single most famous complaint about this engine,
and it deserves to be explained properly, rather than laughed at.
The crankshaft pulley is at the bottom rear of a flat engine,
turning in a vertical plane.
The cooling fan sits on top of the engine, turning in a horizontal plane.
The belt therefore has to run off a vertical pulley, take a 90-degree twist,
wrap the flat-mounted fan pulley, and take another 90-degree twist to get back,
guided around the corners by idler pulleys.
Every engineer who ever saw that drawing said the same thing.
It will come off.
And it did.
Early on, at high rpm, where the fan is absorbing the most power
and the belt is under the most load in the least stable geometry,
Chevrolet's fixes arrived as a running sequence rather than a single redesign.
Deeper pulley grooves to hold the belt in the V, a deeper groove in the idler,
and guides to keep the belt from walking out when it whipped.
Most of that came through during 1960.
And on the 29th of November, 1963,
the belt guard was directed into the top shroud itself,
the date this chapter's evidence has been walking toward.
The pressed steel pan picked up yet another job.
That is not the only time the shroud absorbed a problem.
In December 1959, before most 1960 Corvairs had been sold,
technical service bulletin DR-424 went out to dealers announcing a running change
to prevent carburetor icing by drawing warm air to the carburetors.
Carburetor icing happens because fuel evaporating in the Venturi
chills the incoming air below freezing.
On a damp 40-degree morning,
ice forms inside the throat and the engine stumbles or dies.
On a water-cooled car, you take heat off the coolant.
On a Corvair, there is no coolant,
so the only warm thing available is the cooling air itself.
And the warm air had to come through, and from, that shroud.
Dealers were doing this work by hand, car by car, on cars already sold.
Enthusiast archives that have catalogued the top shroud across 10 years of production
find a part that never stopped changing.
New carburetor notches, revised ducting, heater provisions, added guards,
some of them regular production changes,
some of them temporary fixes pushed out through service bulletins and assembly manual instructions.
Here is the thing to take from that.
Every one of those fixes was cheap.
Changing a stamping die, or adding a small pressed guard,
or cutting a new notch is a small money change at Chevrolet's scale.
Nine years of continuous correction to the airpath cost almost nothing per car
because of what kind of part it was.
Which raises the question this chapter cannot answer on its own.
Why, when Porsche's own engineers had one of these cars in Stuttgart to look at,
did none of that transfer?
Ferry Porsche wanted a Corvair and he got one of the first built.
He arranged it through Huschka von Hanstein,
Porsche's racing manager and public relations director.
A man whose job was to know everybody and get anything.
So within months of the car's October 1959 launch,
one of the earliest Corvairs to leave an American factory
was being driven around Stuttgart by the head of Porsche.
It was not the only one.
Dan Gurney, who drove for Porsche in Formula One in 1961 and 1962,
brought one over himself, by the account that survives,
to show what the new world could do with a car.
And later, Porsche used Corvair Lakewoods, the station wagon version,
as running test vehicles while developing its first flat sixes.
Read that again because it is the most uncomfortable fact in this story
for anyone who thinks of these two companies as opposites.
While Porsche was developing the engine that would define it for the next four decades,
some of the test mileage went on underneath an American Chevrolet estate car.
That was not deference.
It was good engineering practice and it is exactly what a serious development department does.
Get hold of the only other mass-produced rear-engined air-cooled six in the world,
put it into daily use and find out what it actually does across a winter.
And they had plenty to work with.
The Corvair's failings were not hidden from anyone in Stuttgart.
Porsche people could see the front anti-roll bar had been left off to save money.
They could see the rear weight bias had been allowed to run higher
than Ferdinand Porsche would ever have permitted in his own designs.
This was a house that had been rigorous about that specific parameter for 30 years.
They could see the steering was deliberately slowed
so that ordinary drivers would not provoke the car.
They could see that the whole handling package
depended on the owner maintaining sharply different pressures front and rear,
which is a reasonable engineering request in Germany and an unreasonable one in a country
where tire pressures were whatever the last attendant set them to.
None of those observations was wrong.
Every one of them was correct.
And several of them were the same faults American engineers were already arguing about internally.
And a Stuttgart magazine, Automotor und Sport,
road-tested the car and called it Europe's lost opportunity.
A judgment that says the German industry press understood perfectly well
how radical a thing Chevrolet had just done
and regretted that no European company had done it first.
So Stuttgart had the car, drove the car, tested against the car and respected the car.
The question is what they took from the inside of it.
And the record leaves us one specific dated physical clue in a crankshaft.
The first prototype engine for the future 911,
Lennart Jentschke's Type 745 of 1961, had a four-bearing crankshaft.
So did the Corvair.
So did Continental's tank six.
So did most aircraft flat sixes of the period.
Nobody can prove causation from that.
And the source that records the coincidence is careful to say so.
It would be invidious to claim Chevrolet's effort influenced Porsche.
But the sequence is documented and the dates are firm.
Chevrolet's engine with four mains went on sale in October 1959.
Ferry Porsche bought one.
Jentschke's six-cylinder prototype of 1961 had four mains.
And for the final design, Porsche went to seven main bearings instead.
A decision the record credits with securing the long life of the Porsche flat six.
That is worth understanding physically.
A main bearing is where the crankshaft is supported in the crankcase.
Four bearings on a six-cylinder crank means some journals are unsupported over a longer span.
So the crank can flex like a diving board between supports.
And flex is what kills bearings and cracks cranks at high RPM.
Seven means every crankpin has a support on both sides.
It is stiffer, quieter, and it will take far more revs for far longer.
It is also considerably more expensive.
More machining operations, tighter alignment tolerances across a longer set of bores,
and a heavier crankcase.
Porsche chose the expensive answer and was right to.
Because Porsche was building an engine that would be revved hard for 33 years.
Chevrolet chose four and was right to.
Because Chevrolet was building an engine for a family sedan redlined at 5,300
and priced against a Ford Falcon.
Here is where the competent observer goes wrong.
And it is not a technical failure at all.
When an engineer looks at another company's product,
he reads every choice as an engineering choice.
As an answer to a physics question made by someone like himself in a room like his.
It is the most natural thing in the world and it is usually right.
Four bearings instead of seven really is an engineering choice about stiffness against cost.
But a great many of the Corvair's most important choices were not engineering choices at all.
They were tooling choices, decisions about what kind of machine,
already installed in an American factory for another purpose,
could produce this part by the hundred thousand.
You cannot see a tooling choice by disassembling the part.
The finished component gives you its dimensions, its material, and its tolerances.
It does not tell you what press made it, how many seconds it took, or what it cost.
A visitor holding a Corvair shroud learns everything about what it does
and nothing about why it was possible.
Which means the question Stuttgart was asking was the wrong question
and it is worth stating plainly what the right one was.
The question on the table in Stuttgart was,
how do you cool 145 cubic inches of air-cooled six in a family sedan?
That is a good question.
It is answerable.
Porsche answered a harder version of it with the 901,
six individual finned heads, viral cylinders from Mahi,
a 9.8-inch axial fan in a fiberglass duct at 1.3 times crank speed,
the alternator inside the fan hub,
and the answer worked for 33 years in production.
It is also the wrong question,
and being right about it got nobody anywhere.
The real question was never whether air could cool that engine.
Chevrolet had proved it could in a borrowed Porsche at Milford in late 1957
and up Pike's Peak in early 1958 before the first Corvair prototype existed.
The real question was this,
who in the world could manufacture that air path to a consistent shape
a quarter of a million times a year for the price of a family car
and then change it eight or ten times over a decade
without anybody noticing the cost?
That is not a thermodynamics question.
That is a question about presses,
dyes and the industrial base standing behind them,
and it has a completely different set of right answers
from the thermodynamics question
because the companies capable of answering it
are a different list of companies entirely.
Porsche's engineers were not failing to see something subtle about heat transfer.
They understood heat transfer better than Chevrolet did.
What they could not see from inside a company
that built cars in the low tens of thousands
was that the decisive component was not the clever part.
It was the dumb part,
and the dumb part was only cheap
if you already owned the industrial capacity to make it,
which Porsche did not and never would.
This is also where the other common version of this story goes wrong,
and it deserves one honest clause rather than a chapter.
The Corvair was not ended by a verdict on its engineering.
Nader's book landed in the autumn of 1965.
Sales fell from over 237,000 in 1965
to under 104,000 in 1966 and under 30,000 for 1967.
The corporation had already frozen the Corvair's engineering development
in April 1965 before the book,
and the Camaro was approved to fight the Mustang.
As for the later exoneration people cite,
Nitze's report, catalog number PB211015, published July 1972,
it is a real document with real limits.
Its test program, following a letter of the 17th of March 1971,
evaluated the cars in fully loaded condition.
Reasonable people read it differently,
and anyone who tells you it settles everything has not read it.
None of that touches the cooling system.
The cooling system never failed commercially,
never failed legally,
and never failed in the field in a way that ended the car.
It just took 26 horsepower at 6,000 rpm,
and it worked 1,786,243 times.
The answer teased at the start is one word long.
Here it is.
The answer is stamping capacity.
Not the engine, not the fan, not metallurgy,
and not anything Ed Cole or Al Colby invented.
What allowed General Motors to sell air-cooled six-cylinder cars
to Americans at family car prices 1,786,243 times
was that Chevrolet already owned
an enormous sheet metal stamping infrastructure,
dies, presses, press operators,
and the logistics to move pressed panels,
built for an entirely unrelated reason,
which was making car bodies.
The Corvair's bodies came out of Willow Run,
a plant erected to build heavy bombers
A company that stamps a quarter of a million roof panels a year
can stamp a quarter of a million cooling shrouds a year
for something close to the cost of the steel.
And the second half of the answer,
without which the first half means nothing,
Chevrolet was willing to pay for cooling in horsepower
rather than in money.
Approximately 26 of them at 6,000 revolutions per minute,
rising steeply above that,
taken off a car whose base engine was rated at 80 gross.
Porsche could not make that trade.
Porsche sold the horsepower figure.
Ferry Porsche's 130 din was the whole specification,
so every unit the fan consumed came off the product.
Chevrolet sold transportation
and could hand the fan whatever it wanted.
That is why it could not be said out loud.
No marketing department on earth
announces that its cooling system consumes a quarter of the engine,
and no engineering department advertises
that its most decisive component is a pressed steel lid
that a stamping plant runs off between body panels.
The story had to be told as innovation,
because the truth was infrastructure.
Now go back to the two dates that were left standing earlier.
The 29th of November, 1963,
the belt guard directed into the top shroud.
Chevrolet had a four-year-old, world-famous problem.
The belt that turned 90 degrees twice
and came off above 4,000 rpm,
and the fix was another bend in a piece of sheet steel.
Not a redesign of the drive,
not a new accessory layout,
not a gear train,
a guard,
a die change.
In 1964, the cooling fan itself changed from steel to magnesium,
a fan of around 11 inches with 11 straight blades,
replacing the earlier steel fans with their 16 or 24 angled blades.
Magnesium is roughly a third lighter than aluminum by volume
and far lighter than steel.
A lighter fan has less rotating inertia,
which means less shock load into that twisted belt
every time the throttle snaps open
and less energy stored in the spinning mass.
The belt problem and the magnesium fan are the same problem,
attacked from both ends five years apart,
both of them cheap.
Say the disputed figures honestly,
because this audience will check them.
The engine's dry weight is given as 366 pounds
in one widely cited specification,
while owners weighing real engines
report figures in the low 300s and below,
around 312 pounds for a 1966 110 horsepower unit in one account,
lower still for stripped examples,
and one restoration source puts the bare engine at approximately 300.
The band is real,
and the difference is mostly about what is counted,
shrouding,
clutch,
flywheel,
carburetors,
oil.
Chevrolet's original internal target
was substantially below what it built.
The accounts do not agree,
and nobody should pretend they do.
None of that is the point.
The point is that the heaviest engineering in this engine
went into things nobody could see,
and the decisive engineering
went into something everybody could see
and nobody looked at.
There is one more piece of evidence,
and it is the most damning kind,
because it comes from inside the building.
Frank Winchell did not think the Airpath was good enough either.
Winchell ran Chevrolet research and development
through the Corvair years,
and he was the Corvair's most serious internal advocate,
the man who ran the experimental work,
who put prototypes through handling limits
most engineers never approach,
and who was still developing the concept
when the corporation had lost interest.
Under Al Kolb's engine organization,
with engineers including Joe Birch,
Len Kutkus,
and Jerry Misterlick,
Chevrolet worked on a modular air-cooled engine family
intended to succeed the Turbo Air 6.
And in that engine,
the single horizontal cooling fan was gone.
In its place,
three fans,
mounted vertically
on a common shaft above the cylinder banks,
blowing air across the engine.
That is Chevrolet's own verdict
on its own Airpath,
written in hardware.
The best engineers in the building,
given a clean sheet
and no obligation to reuse the 1958 layout,
did not keep the flat centrifugal blower
Kolb had carried over from the Continental tank engine.
They split it into three and stood them up.
Everything else the record shows them fighting in that program,
heat distortion of early alloy cylinder barrels among it,
points the same way.
The Airpath was the hard part,
permanently,
for nine years and beyond.
None of it reached a customer.
The Corvair's engineering development
was frozen in April 1965.
The third-generation car,
XP849,
badged Corvair 2,
reached full-size model stage by June 1965,
was shelved,
was revived by Kolb in June 1966
under corporate styling and central engineering staff,
and was shelved again in June 1967
in favor of a front-engine coupe concept.
A last rear-engine study,
XP892,
was planned around a water-cooled 4,
and Kolb killed it at the end of June 1968.
He turned to XP887 instead,
which arrived in 1970
as the Chevrolet Vega,
front-engined,
water-cooled,
aluminum-blocked,
now the correction to the story that survived.
The version in circulation
says the Corvair was a flawed idea
that a book destroyed.
The primary record says something less dramatic
and more interesting.
The cooling system
was never the thing that failed.
It failed nowhere.
The engine that was supposedly too radical
outlived its own reputation
by shipping 1,786,243 times,
and the last car came off the line
on the 14th of May, 1969,
a gold two-door hardtop.
Kohl himself,
after the 1973 oil embargo,
said that had it still been in production,
Chevrolet could have sold
every one it could build.
That version survives
because a court fight
and a Senate hearing
make a better story than a stamping plant,
and because the man who told it best
was not an engineer.
If this is the version you want more of,
the one built out of part numbers,
directed order dates,
and the things engineers actually wrote down,
rather than the version
everyone has repeated for 60 years,
then give it a like,
because that is the only way
it reaches the people
who would rather have it right
than have it familiar.
And subscribe,
because this one
leaves a real question standing.
Chevrolet had a modular air-cooled engine
with three vertical fans on the bench,
and it chose the Vega instead.
Why the corporation abandoned air-cooling
at the exact moment
it had finally learned to do it properly
is the next file,
and it runs straight through
XP 849's cancellation in June, 1967.
Last, the people.
Edward N.
Kohl, who flew 300 hours
behind a Continental
and hit a car
behind Australian paperwork.
Adalbert Kolb,
who filed the layout patent
on the last day of 1958.
Bob Benzinger,
who put a carburettor
on each bank
and explained exactly why.
Frank Winchell,
who kept working on it
after the corporation stopped.
Klaus von Recker,
who could not make the number
and left.
Hans Tamala,
who took the project on.
Leonard Jenschke,
whose four-bearing prototype
came first.
Hans Mesger,
who drew the engine
that lasted 33 years.
And the men and women at Messina,
who poured the aluminium,
at Tonawanda,
who assembled it,
and at Willow Run,
who pressed the steel.
None of them were components.
They were people,
working under constraints
they did not choose,
and every one of them
got something right.