It began, as good engineering often does, with a complaint. In 1922 the Treaty of Versailles had forbidden Germany from building aircraft engines, and BMW — a company that until then had made almost nothing else — was reduced to air brakes, farm machinery and office furniture. When Franz Josef Popp asked his design director to assess the Helios, a motorcycle the firm had inherited through merger, Max Friz reported back that the best thing to do with it was throw it in the nearest lake. Its engine sat with the crankshaft fore and aft, and the rear cylinder cooked itself in the shadow of the front one.

Friz's answer was not to improve the cooling. It was to turn the engine sideways so that both cylinders stood out in the airstream, where nothing could hide from the wind, and to run the drive straight back through a shaft rather than a chain. The R32 that resulted in 1923 made 8.5 horsepower from 494 cc and would do about sixty miles an hour. Only three thousand or so were built. But the shape of it — a flat twin with its heads in the breeze and a driveshaft down the right side — is still the shape of a BMW boxer a century later. Very few design decisions survive their designer. This one survived the century.

The engine in your workshop is a direct descendant, and it arrived with a change of address. Through the sixties, car production in Munich swelled until the motorcycles had to go somewhere else; in 1969 BMW moved the whole operation to Berlin-Spandau, into a plant that had once built radial aircraft engines for Siemens. What came out of it that year was the /5 — the R50/5, R60/5 and R75/5 — an entirely new machine built on a modular principle from the suspension to the flat twin itself. It sold. Production ran five-fold in three years, and by the time the /5 ended in July 1973 nearly sixty-nine thousand had left the line.

Then the decade the platform is remembered for. In 1973 the engine grew for the first time and gave us the R90/6 and the R90S, that smoked-orange half-faired thing that made touring bikes look fast. The /6 of 1974 brought a front disc and the five-speed box. In 1976 the R100RS arrived as the first production motorcycle in the world to wear a full frame-mounted fairing, developed in a wind tunnel, and the R45 and R65 filled in beneath it. Annual production passed thirty thousand for the first time in 1977. This is the high-water mark of the airhead: the range at its widest, the engineering at its most confident.

And in 1980, almost as an afterthought, the machine that outlived all of them. The R80 G/S was the world's first production motorcycle with a single-sided swingarm, and on paper it was absurd — a 797 cc shaft-drive twin weighing four hundred pounds, entered into a sport where everyone else rode light singles. In 1981 Hubert Auriol took it to overall victory in the Paris–Dakar, the first ever for a multi-cylinder motorcycle. He won again in 1983; Gaston Rahier, all five foot four of him, won in 1984 and 1985. The photographs of that enormous tank and those cylinder heads jutting into the Sahara did more for the boxer than any brochure ever printed, and every adventure bike sold since owes it a debt.

The end came quietly and from inside the family. The water-cooled K100 landed in 1983, and in 1993 — seventy years after the R32 — the R1100RS introduced a boxer with oil cooling and four valves per head. Only then did anyone need the word airhead: it is a retronym, coined backwards to describe what had just been superseded. The last of them, the R100R and its Mystik variant, trickled out of Spandau into 1996. Then the line that Max Friz drew in December 1922 finally stopped moving.

Which brings us to the bench. What you have is not a vintage motorcycle in the fragile sense — it is a mature industrial design that was refined for seventy years and then discontinued while it was still good. The tolerances in this book are wide because the engine was built to be serviced by hand, at the roadside, by someone who owned a feeler gauge and a decent memory. Nothing on it is hidden. Nothing on it requires a laptop. It will tell you what is wrong if you know how to ask, and everything after this chapter is a way of asking.

Twenty-five years of production, and the differences between one end and the other are larger than newcomers expect. A 1972 /5 and a 1993 R100R share an engine layout and very little else — different brakes, different suspension, different ignition, different character.

What follows is the shape of the range, the years the long-standing names in this hobby single out, and the specific faults worth knowing before you hand over money. One caution before any of it: as one MOA member put it, there are no trouble-free airhead years. Every model has its warts — the question is only which ones you are willing to live with.

Everything else in this book answers a question you already know to ask. This chapter is for the questions you do not — the order to do things in, the habits that save a weekend, and the mistakes that are common enough to have names.

None of it is difficult. Airheads were built to be maintained by their owners, with hand tools, at the roadside if necessary. What catches people is sequence: the right job done in the wrong order wastes a fortnight and teaches you nothing.

Most of an airhead comes apart with ordinary metric tools and patience. What stops the job is never the socket set — it is the one purpose-made thing you do not own, discovered at the point where the engine is already in pieces.

This chapter separates the tools you need before you start from the specials you can buy when the job demands them. Prices drift; treat them as order-of-magnitude. The one rule worth stating plainly: buy the tool before you improvise, because on a forty-year-old casting the improvised version usually costs more than the tool.

There is no single factory schedule that spans 1969 to 1995. BMW's published intervals changed with the models, and the only one reproduced verbatim and verifiable today is the /5 booklet — a two-tier 3,000-mile minor and 6,000-mile major service. Everything after that has to be assembled from the factory literature that survives, the independent authorities, and five decades of what actually went wrong.

So this chapter gives you both columns. Where BMW and Robert Fleischer disagree — and they disagree often — you get both figures and the reason. Where the factory schedule is simply silent on something that destroys airheads, that is called out, because the most consequential item on the whole ladder does not appear in any factory service booklet at all: the transmission input splines.

Two structural warnings. First, several of the items below are governed by time, not mileage — brake fluid, fork oil, fuel line, tyres, battery — and those are the ones a low-mileage bike misses entirely. Second, every interval here assumes moderate use. Sidecar, trailer, heat, dust or a diet of short trips moves most of them earlier, and the ladder says where.

Every number in this chapter is a compromise between two failures. Too little and the joint works loose under vibration that never stops; too much and you stretch a stud, pull a thread out of the forty-year-old aluminium crankcase the long cylinder studs are threaded into, or warp a head that was flat when you started. BMW published figures for all of it. Some of those figures are wrong.

That is not a slur — it is documented. In March 1982 BMW issued a service bulletin telling its own dealers to stop trusting the foot-pound conversions printed in the workshop manuals, because they contained errors. Those errors were copied forward into Clymer and Haynes, and they are still being quoted in forum threads today. Convert from Newton-metres yourself, every time.

Where this reference disagrees with the factory book, it says so and explains why. The values marked with a warning are the ones airheads do not forgive.

An airhead tells you how it is doing through its valve gear. The tick that turns into a tap, the exhaust note that goes flat on one side, the oil smoke that appears only on overrun — all of it starts up here, under two aluminium covers a spanner can have off in a minute.

Which is the good news: this is the most accessible serious maintenance on the bike, and the most rewarding. A cold engine, two feeler blades and twenty minutes buys you the difference between an engine that runs and one that runs right.

Clearances changed by model and by year, and the /7 figures are tighter than the /5 numbers most people quote from memory. Find your bike in the first table before you touch anything.

You will open the bottom end perhaps once in your ownership, and when you do, everything depends on numbers you cannot judge by feel. Bearing clearance, end float, rod side play — these are dial-indicator and micrometer territory, and the difference between right and wrong is measured in hundredths of a millimetre.

Two figures here earn special attention. Crankshaft end float is the quiet cause of a dozen symptoms owners misdiagnose for years: the grabby clutch, the awkward shift when hot, the idle that changes when you pull the lever. And on 1981-and-later engines, main bearings became a selective colour-coded fit — a genuine trap for anyone working from an older manual.

In late 1980 BMW changed what an airhead cylinder fundamentally is. Before: a cast-iron liner in a finned aluminium jacket, honeable, reboreable, forgiving. After: an aluminium barrel with a hard plating bonded directly to the bore — lighter, better at shedding heat, almost immune to wear, and impossible to rebore.

If your bike is post-1980, that single sentence governs everything. A worn barrel is not a machine-shop problem; it is a new-parts problem, or a specialist replating job. It also changes your ring gaps, your piston clearance, and which honing you must never let anyone perform on it.

Everything below is split accordingly. Check your build date first.

The broad strokes: what the engine is, how hard it may be worked, and the compression numbers that tell you whether the rest of this book applies to a healthy motor or a tired one.

Worth knowing before you read your gauge: American-market 1000 cc bikes from 1980 were dropped to 8.2:1 compression to meet emissions (other markets and later models differ — check yours). Combined with the lean jetting of the same era, that shapes almost every decision you will make later about fuel, ignition and cylinder kits.

Constant-depression carburettors confuse people because nothing moves when you expect it to. Twist the throttle and the slide does not obey your wrist — it obeys the vacuum in the venturi, rising only as fast as the engine can actually use the air. That is why an airhead pulls so cleanly from low revs, and why the usual tuning instincts lead you astray.

The practical consequence is this: when the mid-range goes soft, rich or hesitant, almost nobody needs a bigger main jet. They need a needle and a needle jet, because those two wear together and take the whole middle of the throttle range with them.

And identify your carburettors by the number stamped on the flange. Not by what model the bike is. Bodies were shared across models; jetting was not.

The airhead gearbox has a reputation it half deserves. It is not fragile, but it is unforgiving of assembly error, and it hides its complaints behind symptoms that sound like clutch problems.

Shimming is the craft at the centre of it. Three shafts, each wanting five to ten hundredths of a millimetre of end play — measured cold, because the aluminium case grows faster than the steel inside it. The factory method is elegant once you see it: shim every shaft to zero, then let the crush of a new gasket donate exactly the running clearance you need. Fit a gasket of unknown thickness and you have silently changed the specification.

Further back, the final drive ratio stamped on the case tells you more about how your bike will feel at seventy than any other single number on the machine.

This is the chapter that saves gearboxes. Everything else in this book is a number you look up once; this is a job you have to keep doing, on a clock nobody at BMW ever wrote down. The factory service schedule does not mention the transmission input splines at all. They are, nevertheless, the single most consequential scheduled item on an airhead — and the failure they produce is expensive, unheralded, and entirely preventable.

Be careful with the popular framing. There is no canonical "five spline points" list for airheads; that list belongs to K-bikes and later machines. A twin-shock airhead has three greased spline interfaces plus one that lives in oil. A Monolever or Paralever has fewer still — no rear wheel splines at all. The table below separates the points you actually service from the ones you only ever touch during assembly, because conflating them is how people end up disassembling something that needed nothing.

The deeper disagreement is worth stating up front, because it changes what you conclude when you find wear. One school holds that dry, rusting splines are the cause: grease them on schedule and they last. The other observes failures at 20,000 and 30,000 miles with the grease still intact, and points instead at misalignment — a distorted clutch disc, a rocking hub, a bent housing. Both are documented, by serious people. If you find heavy wear on a well-greased spline, do not simply re-grease it and reset the clock.

Most airhead electrical mysteries are not mysteries. They are a handful of well-documented design consequences that surprise each new owner in turn.

The best example: on a standard Bosch system, the charge warning lamp is not merely an indicator. Its current is what wakes the rotor field. Burn out that bulb — or fit a modern LED that draws almost nothing — and the alternator stops charging entirely, while the dashboard sits there looking perfectly content. A three-cent resistor solves it forever.

The second: coil primary resistance must match the ignition driving it. Points systems want roughly three ohms total across the pair; the electronic systems from 1981 want substantially less — around half, and the exact figure depends on the ignition and coil generation, not on a fixed ratio. Take the number from the table below, or from the ignition maker if the system is aftermarket. Mix them and you will either cook a control unit or wonder for months why the spark is weak.

Four separate oil compartments, each with its own fluid, its own interval and its own drain plug hiding somewhere unhelpful. Get familiar with all four and the bike will outlive you.

The specification most often quoted wrongly is fork oil. The older 235 cc figure is still repeated everywhere, but 1981 and later forks take less, are measured by a rod-drop level rather than by volume alone, and are notably fussy about foaming — which means the fluid you choose matters more on a later bike than an early one.

One habit worth adopting from this chapter: airheads dislike long idling. At a hot idle, oil delivery to the timing chain is marginal and bearing films thin out. Ride it or shut it off.

There is a version of this hobby where originality is the point, and this chapter is not for it. This is for the rest of us — the people who want a motorcycle of this age that starts on a cold morning, charges at a traffic light, stops in the wet and does not electrocute itself when the loom finally gives up.

The good news is that the airhead aftermarket is unusually mature. Charging systems that make three times the original output, ignition that never needs a feeler gauge again, control electronics that replace an entire relay board with something the size of a cigarette packet, cylinder kits that answer the no-rebore problem outright.

Each entry below notes what it actually buys you and, more importantly, what else it forces you to change. Nothing on a motorcycle gets upgraded alone.

Finish work is where good builds are won and lost, because it is the part everyone sees and the part that fails first when the preparation is rushed. Every coating in this chapter depends more on what happens before it is applied than on the product itself.

The chapter also contains an honest disappointment. If you have heat-coloured a set of headers and want to seal that gradient under a clear coat, there is currently no clear rated for the temperatures at a header flange. The options are a compromise, a traditional oiled finish, or an opaque coating that hides the very thing you made. Better to know that before you spray than after.

Three decades after the last of these bikes left Berlin, parts availability remains extraordinary — provided you search the way the trade searches. That means by part number, always, sourced from the fiche for your build date rather than your model year.

What follows is a map of the whole supply landscape: the dealers who still stock genuine parts, the specialists whose aftermarket versions improve on the originals, the breakers who hold what nobody makes any more, and an honest note about which systems are getting hard to feed.

Specifications describe what a motorcycle is. Photographs argue about what it could be.

Every torque figure and clearance in this book exists in service of something, and this is the something — more than a thousand airheads that people cut apart and put back together, indexed by model, style and builder. The photographs and the words stay with the magazines and blogs that made them; what is here is the map to them, filed the way an airhead owner actually looks: by what you ride.

Some model years sit comfortably in the middle of a production run. 1981 is not one of them.

The autumn of 1980 brought the largest single redesign in airhead history — new crankcases, plated cylinder bores, electronic ignition, revised bearing fits and a scattering of changed torque figures. A bike built weeks either side of that line needs different parts and different numbers.

This chapter gathers every boundary in one place, so the question you ask a parts supplier is never "what year?" but "before or after September 1980?"

A reference is only worth the sources behind it, and the sources for these motorcycles do not always agree.

Where they conflict, this book takes a position and shows its working rather than quietly picking a number and hoping. The head-nut torque disagreement documented below is the clearest example: the factory-derived figure and the most respected independent authority differ by nearly twenty percent, on a fastener that will happily pull a cylinder stud out of the aluminium crankcase. (The long through-studs thread into the case, not into the head — a pulled thread there is a bottom-end repair, not a head swap.)

Verify anything critical against your own manual before final assembly. Steel does not forgive.