A Predictable and Measured Failure
The story of composite aircraft structural development
A Predictable and Measured Failure
The history of flight is told as a climb toward stronger materials and cleverer machines. Run it as a failure analysis and a different story appears — one nerve, from Kitty Hawk to the Dreamliner. Aviation has only ever been the fight to make the invisible predictable, and a material crosses into the sky at exactly the speed its failure can be seen — slowest of all where there is the most to lose.
Watch a bird from a chair and you see it flapping. That is the whole illusion — that flight is propulsion, two wings beating against gravity. It is what the eye reports, and for a century it is what men built toward: bigger wings to flap, more horsepower to flap them. Every one of those machines stayed on the ground.
What the eye cannot see is what the bird is actually doing. Between the beats it is trimming — twisting the trailing edge of a wing, curling a tip against a gust, correcting with the tail a thousand times a second. The bird is not mostly flapping. It is mostly balancing, and the balancing is invisible at a distance. The seen part, the flapping, is the decoy. The unseen part, the continuous correction, is the airplane.
That distinction is the whole of aviation, and everything that follows is one argument: the machines that fly, and the materials that carry them, advance only when someone drags the invisible thing — the correction you can’t watch, the failure you can’t see — onto a schedule you can trust. The achievement was never lift, or strength. The achievement was the eyes.
Two men who read the invisible
The Wright brothers won because they went looking for the part of the bird nobody could see. Wilbur watched buzzards over Dayton and noticed they did not beat their way back to level in a gust — they twisted their wingtips and rolled upright. Wing warping came from reading a correction the eye almost misses. While the rest of the field chased the visible problem, power, the Wrights chased the invisible one, control, and then did the thing that actually mattered: they made the correction predictable. Warp plus a coordinated rudder gave a pilot a known, repeatable response — the first working three-axis control. Not a lucky airplane. A controllable one.
They held the visible keys too, and earned each by measurement rather than hope. They proved their wing by kiting it and gliding it, three seasons in the sand — and when Lilienthal’s published lift tables betrayed them, they built a wind tunnel and generated their own numbers. When no engine maker would meet their power-to-weight, they built a light aluminum-block motor. Then came the part almost everyone forgets: no one had a real theory of the propeller, so they treated the prop as a rotating wing, ran their own airfoil data through it, and carved propellers roughly seventy percent efficient in 1903 — a figure that stood for decades. Proven lift, working control, buildable power: three keys, on one bench, in two men’s hands.
Set them against the man who should have won. Samuel Langley had the Smithsonian, a War Department contract, and money the bicycle mechanics could only dream of. He ran his program the way an institution runs one — from the top, by proxy, the Secretary presiding while Charles Manly built the engine and other men did the flying. He had lift, and he had power. What he never had was control, because control is the one thing that cannot be delegated: it lives in the seat, learned only by riding the machine as it tries to kill you, and Langley was never in the seat. His Aerodrome went into the Potomac twice. He held one key and a fat checkbook; the Wrights held all three in their fingers. First flight did not go to the richest or the first to try. It went to the only shop where proven lift, working control, and buildable power sat on the same bench. Convergence is not a checkbook. It is every key in the same hands, in the same season.
The material that tells you when it will die
The Wrights built their airplane out of a composite, though no one used the word. Wood is cellulose fiber in a lignin matrix — stiff along the grain, weak across it — and they ran the grain of every spar on purpose, the way a layup technician orients carbon ply today. The first airplane understood that structure has a direction. Then aviation spent seventy years forgetting it, in metal.
The standard story says aluminum won because it was better. It wasn’t, not on strength — oriented fiber beats it pound for pound. Aluminum won for a subtler reason than even its defenders usually name. It was not merely that aluminum fails visibly. It was that aluminum fails predictably and measurably: a known fatigue curve, a crack that grows at a calculable rate, a yield you can compute and count on. Metal tells you when it will die, in numbers, in advance. That predictability is the thing certification is built to reward — and it is exactly what early composites lacked. Smooth, moldable, stiff, strong, and mute about their own death: they would hold, and hold, and hold, and then let go without a measured warning. The aluminum century was not a triumph of metal. It was a seventy-year refusal to build anything that failed on a schedule no one could yet read.
So the real history of the composite is not “it came back.” It is the long project of teaching a mute material to fail on a schedule — of making its failure predictable and measured, the way metal’s already was. Every airplane that follows is a station in that one project.
The edge, and the ladder
There are two tiers to this history, and blurring them is what makes people call it a tale of bold outsiders. There is the tax-free edge — experimental articles, research prototypes, homebuilt kits, where a single airframe and a volunteer pilot are all there is to lose, so the material could go all-composite at once, and did. And there is the certified ladder, where a type certificate, a company, and a stranger’s life are on the line, and the material could only climb one measured, over-built, inspected step at a time. Which tier an airplane sits on is not decided by courage. It is decided by one number: how much a hidden failure would cost.
Start at the edge. The Marvel proved it where no one had to certify it. In 1964 the U.S. Army gave Mississippi State University’s Raspet Flight Research Laboratory a contract for a short-takeoff research aircraft, the XV-11A Marvel — fiberglass essentially throughout, first flown in 1966. That makes it the first all-composite airplane, full stop, and it flew where the blindness tax could not reach it: a research contract, an experimental article, an airframe the government could afford to lose. The material always showed up first at that edge, because there was almost nothing at risk if the failure no one could see arrived.
The old tool could find the flaw but never read its future. For generations the field test for a composite was a coin. Tap a quarter along the skin and a trained ear hears it — a clean ring over sound laminate, a dull thud over a delamination or a debond. It works, and it costs nothing. But it is a historical metric as limited as it is honest: it marks a location and nothing more — no depth, no extent, no count of the cycles left before the part lets go. It finds the spot and goes silent on everything that decides whether the airplane is safe to fly. And that is the exact line between the two tiers. At the edge, detection is enough: find it, cut it out, lay it up again. Certification asks a different question. It does not ask is something wrong; it asks how deep, how wide, how fast is it growing, and how long until it matters — a bounded, measured prognosis a coin can never give. The quarter offers a flinch. The type certificate demands a forecast. That gap — detection turned into prognosis — is the whole distance the composite had to travel to be trusted over your head, and every rung of the ladder is paying to cross it.
Windecker paid the tax, and paid it again in the market. In December 1969 the FAA certified the Windecker Eagle — the first certified all-composite airplane, built by Leo and Fairfax Windecker, two dentists who were fluent in the material and innocent of every reason aviation feared it. The certification record is the blindness tax in its crudest form. With no way to inspect a composite structure the way it inspected metal, the FAA made them build the Eagle twenty percent stronger than the aluminum equivalent and then spin-test it nearly to death — a prototype was lost in the attempt, the pilot bailing out with inspectors watching. When you cannot predict the failure, you buy confidence by brute force: overbuild it, then test it past reason. And the airplane still died. A skeleton crew hand-built about one Eagle a year; nine were completed; the company folded in 1970 when its backers walked. A seed in ground that was not ready.
The Starship was dragged into the cage and killed there. Burt Rutan is the arch-outsider, and the tell is that he almost never certified anything. His genius lived at the tax-free edge — homebuilt kits sold as plans in the experimental category, one-off research and record aircraft — because the edge is where an outsider can push the envelope without paying for the invisibility of the failure. The Starship is the single time a Rutan concept was taken all the way into certified production, and Beech did the certifying, not him. The tax deformed it to death: the FAA forced flaps, which forced a swinging canard, which forced weight, until empty weight ballooned some 2,400 pounds over target, crossed the type-rating threshold, and priced the airplane out of its own market. Beech built fifty-three, sold fewer than a dozen, then bought back and burned nearly every one. The edge is where you can be right and survive. Certification, too early, is where being right gets you killed.
Line the pioneers up and the pattern has a body count: every all-composite airplane that crossed into certified production died in the market — the Eagle, the Starship, and later the Rutan-derived Adam A500, seven delivered before bankruptcy. Meanwhile the same material thrived wherever the tax couldn’t reach: Rutan sold thousands of homebuilt plans; the Marvel flew on Army money. Many seeds were planted; few took root. A seed needs three things in the same season — the material mature enough to trust, a builder production-ready enough to make it at rate, and a market ready to buy that airplane — and almost no one ever had all three keys at once. Windecker had a material and no organization. The Starship had Beech’s organization and neither a mature material nor a market. They were not foolish. They were under-resourced against a three-lock door.
The jet step changed the regime. Notice what those pioneers shared besides dying: they all turned propellers. The Marvel, the Eagle, the Starship were the propeller world’s composites — the proving ground. What came next was not the next rung but a different failure environment: the jet step, where composite primary structure had to survive high-altitude turbofan flight. That is the hardest place there is to make a failure predictable. A jet transport cruising near forty thousand feet is a pressure vessel that inflates and deflates tens of thousands of times across its life, and every cycle is an open invitation to the exact fatigue crack a composite hides best — at speeds and loads no Bonanza ever saw. Proving the material in the propeller world was the warm-up. The jet step was the examination. On the certified ladder only two airplanes have ever passed it — but before them, an uncertified reach tried for the same thing and found the boundary, and that reach is where my own thread runs.
One program at the edge set out to learn the whole craft — and I learned it there. Twenty years after the Marvel flew, in the mid-1980s, I was at that same laboratory, and some of the men who had built it taught me to lay up composite by hand — how to orient the cloth, wet it out, chase the resin, read a part you cannot see the inside of. The program had one stated goal: to master, end to end, how to build a full composite airframe. It began the way you learn anything real, by rebuilding something known — a Beechcraft Bonanza whose structural components were re-made in graphite, glass, and Kevlar, with the front-mounted turboprop refitted onto the composite conversion. Then the parts were tested by hand to find out what they would actually do, including pressurizing the composite fuselage in a swimming pool to watch whether it held. That is not a figure of speech for building the eyes. It is building the eyes — a measurement rig improvised out of water, because the material would not tell you where it failed unless you forced it to show you.
Once those techniques were learned and mastered — the airframe built all-composite and flown on an experimental ticket that never led to certification — the last and hardest step tried twin jet engines mounted above the wing. That challenged the new material system directly, at the wing, where the concentrated load ran. And here is my testimony from inside the program, not a line from the record: rather than keep re-working the composite wing to carry it, the wing’s makeup was superseded by the priority that ruled the airplane — flight performance. The production wing became machined aluminum, chosen for the smooth laminar surface that speed demanded, and the all-composite wing was let go. So this airplane sits on neither tier. It is the bridge between them — the program that learned the whole craft, carried it by hand off the uncertified edge toward the certified sky, and then, at the last step, chose performance over the unproven material.
At transport scale, the jet step crossed the line — and is still fighting for it at the spar. These are the two airplanes that passed the examination on the certified ladder, and they carry more exposure than anything else in the sky — a wide-body holds hundreds of lives and bets a corporation on every hull. By the only standard that matters — wing spar, fuselage, wing skins, the three load paths that decide an airplane — both are all-composite main structure. On the A350 the spars, wing covers, centre wing box, and fuselage panels are carbon fibre; on the 787 the wingbox spars and covers are carbon fibre and the fuselage is wound as composite barrels. Metal still holds the ribs, the fittings, the leading edges — but not the spar, the skin, or the pressure vessel. The primary load path is composite. The material has taken the airplane.
But the predictability project is not finished, and you can see exactly where it is still live, because it is live at the most critical structure there is. During the 787’s static testing, the composite wingbox spars buckled. Boeing traced it to an error in the earlier modeling analysis, added brackets to boxes already built, redesigned the ones not yet built, and slipped delivery about six months. The single most important structure on the airplane failed in a way the model did not predict — and the fix was the same ritual as Windecker’s, unchanged in forty years: you cannot yet predict it, so you overbuild it and you test it again. A Boeing engineer named the whole thesis in one breath — the trouble was never that composites aren’t strong; it’s that they are so internally complex they are hard to analyze. That is “predictable and measured failure,” named as the unsolved frontier, by the people standing on it.
And Airbus writes the rest of the argument into its own maintenance manuals. Strike a metal panel hard enough and it dents, permanently, where you can see it. Strike a composite panel and it flexes and springs back to a smooth, undamaged-looking surface — while the damage hides in the core. So the A350 carries added inspection tasks written specifically for those impacts, an ultrasonic hand tool that will release the airplane only if no delamination is found and escalates to full instrumented inspection if any is, and a metallic mesh woven through the skin to give bare carbon the electrical predictability metal had for free. None of that is doubt about the material’s strength. All of it is machinery for making an invisible failure measurable. The material is trusted for primary structure only to the exact degree its failure has been made predictable — and metal still holds every spot where it hasn’t.
The ground finally ready
The material did root, and grow, and become ordinary — just not for any of the people who proved it. The all-composite Cirrus SR22, certified in 1998, became the best-selling airplane in general aviation. Thirty years after Windecker died broke holding the same idea, someone else sold thousands of it, because by then all three keys had arrived in one season: the processes had matured, the company was built production-ready from the first day, and a market finally wanted a fast, modern composite single. Windecker lived to watch it and told an interviewer that aluminum was never a good material to build an airplane out of anyway. He was right the whole time. Being right did not pay. Being able to make the failure measurable — and holding all three keys at once — did. (The Cirrus won for other reasons too, a whole-airframe parachute and a glass panel among them; the material didn’t win alone. It won once it was finally sellable.)
That is how the composite came back — not up a ladder of triumphs, but as a long planting, most of it lost, a little of it green, the buried seeds conditioning the ground the survivor grew in.
The eyes were the achievement
It was never about what flies. It was always about what you can’t see it doing.
You do not have to fly to know the wing makes lift, or to place the stiffness where the load runs — that is the visible part, the part a stress engineer assigns on paper, dry, the way the Wrights ran the grain of a spar. What has to be flown, again and again, is the correction the eye can’t report and the failure the surface won’t show: the gust the bird answers before you see it, the delamination that hides under an intact skin. The seen part yields to calculation. The unseen part yields only to riding the thing until it shows you where it’s blind — or, once the methods finally catch up, to the rigs and the ultrasound that learned to see it for you.
So the whole story is one nerve, told twice. In 1903 two men learned to see the bird’s hidden correction and make it repeatable. For the next century the industry learned to see the flaw hidden in the ply and make its failure predictable and measured. Flight was never conquered by generating lift or by proving strength. It was conquered by taking the thing that hides from measurement and dragging it onto a schedule you can trust — the gust response then, the failure mode now.
And notice who actually did it — because it was not the outsiders, and I say that having been one. The outsiders proved the material was possible, out at the edge where a single airframe and a willing pilot were the whole of what could be lost. But possibility is not adoption. The composite entered the certified sky through the two most conservative institutions in the industry, Boeing and Airbus, doing the least romantic thing imaginable: inserting it one piece at a time across decades — a rudder, then a tailplane, then the empennage, then at last the fuselage and the wing — testing every step, building the inspection regime, keeping metal until each failure could be bounded, bolting brackets onto the spar the instant it misbehaved. That is not rebellion. It is conservatism, and conservatism is what a type certificate rewards. You are not certified for being bold. You are certified for proving the failure is bounded, known, and inspectable.
Which exposes the rate law running under the whole hundred and twenty years: the more a hidden failure would cost you, the slower you adopt the material whose failure you cannot yet see. The experimental article went all-composite overnight, because it risked one aircraft and a volunteer. The certified single risked a company and a stranger’s life, and paid the tax in dead weight and spin tests. The wide-body, with hundreds of lives and two corporations exposed on every flight, moved slowest of all — one inspected component at a time, thirty years from the rudder to the wing. Adoption ran inversely to exposure, and the curve never broke. What looked like the establishment’s timidity was an accurate reading of what it stood to lose. Metal had been adopted freely only because its failure was already measurable, so nothing was hidden. The composite advanced across the airframe at precisely the speed the eyes could be built to see it fail — and not one step faster. The outsiders were the prologue. The measurement was the achievement.
The material was ready the whole time. What took a hundred and twenty years was learning to see it fail. The achievement was never the composite, or the airplane. The achievement was the eyes — and what the eyes were for was to make the unseen thing measurable, so that a mute material could finally be trusted to break on a schedule, in numbers, in advance. A predictable and measured failure. That, and not strength, is what it means to have conquered the air.
Herbert Roberts, P.E. is a licensed professional engineer with 32 years in aviation research and development across two companies, and has spent eight years analyzing accidents for attorneys under his P.E. license.